Digitalboard/reference-ansible
6
1
Fork 0
Code Pull requests Activity Actions

docs(reference-ansible): add docs/ tree and document repo, playbooks, Makefile

Browse source 
Addresses the WKS PoC review (Notion 2026-05-26). All docs in English.
- README: purpose, docs table of contents, annotated repo tree
- docs/getting_started.md: prerequisites (WKS account, OIDC, SSH, VPN) + first deploy
- docs/ansible.md: playbook table, "Running Ansible", service parameters, cheatsheet
- docs/secrets.md: canonical Bao login (moved out of README) + demo defaults
- docs/operations.md: full Makefile reference
- docs/inventories.md: repo layout, topology, standard folder structure, walkthrough
- docs/testing.md: static checks, inventory resolution, smoke test / dry run
- remove ARCHITECTURE.md (architecture docs live externally)

Also includes the gymburgdorf inventory build-out (bookstack, homarr,
opnform, send) and scripts/bao-seed.sh. site.yml keeps a third traefik
play (traefik_servers minus the vagrant _dmz/_backend split) so the demo
inventories still configure their reverse proxy after the rebase onto main.
This commit is contained in:
Simon Bärlocher 2026-05-27 18:08:52 +02:00
parent c67e9aac43
commit 2ba0c07cd3
No known key found for this signature in database
GPG key ID: 63DE20495932047A
24 changed files with 1541 additions and 525 deletions
Download patch file Download diff file Expand all files Collapse all files

4
.gitignore vendored
View file

@ -12,6 +12,10 @@
.snapshots/* .snapshots/*
# Idea # Idea
/.idea/ /.idea/
# macOS
.DS_Store
# Claude Code local settings
.claude/
# Ansible # Ansible
/collections/ansible_collections/ /collections/ansible_collections/
.vagrant/ .vagrant/

453
ARCHITECTURE.md
View file

@ -1,453 +0,0 @@
<!-- markdownlint-disable MD013 MD060 MD051 -->
# Architektur — `reference-ansible`
Dieses Dokument beschreibt die Architektur des Repos `reference-ansible`
und nutzt die Inventory `inventories/demo-gymburgdorf/` als
durchgehendes Beispiel. Es dient sowohl als Onboarding-Doku für neue
Engineers als auch als Referenz beim Aufsetzen weiterer Demo-Mandanten.
> **Demo-only.** Alle Defaults in den Rollen (Passwörter, Tokens,
> RPC-Secrets) sind unsicher und ausschliesslich für Demo-Setups
> gedacht. Siehe [§ 8 — Security und Demo-Only-Defaults](#8-security-und-demo-only-defaults).
**Letzte Aktualisierung:** 2026-05-18 · **Owner:** @sbaerlocher
## Inhalt
- [§ 0 — Glossar](#0-glossar)
- [§ 1 — Repo-Layout und Roles-Herkunft](#1-repo-layout-und-roles-herkunft)
- [§ 2 — Setup und Voraussetzungen](#2-setup-und-voraussetzungen)
- [§ 3 — Variablen-Hierarchie](#3-variablen-hierarchie)
- [§ 4 — Inventory-Topologie (`demo-gymburgdorf`)](#4-inventory-topologie-demo-gymburgdorf)
- [§ 5 — Service-Layout und Variablen-Verortung](#5-service-layout-und-variablen-verortung)
- [§ 6 — Deploy-Flow](#6-deploy-flow)
- [§ 7 — Traefik-Modi (DMZ vs Backend)](#7-traefik-modi-dmz-vs-backend)
- [§ 8 — Security und Demo-Only-Defaults](#8-security-und-demo-only-defaults)
- [§ 9 — Variablen-Cheatsheet](#9-variablen-cheatsheet)
- [§ 10 — Walkthrough: Neuen Demo-Mandanten anlegen](#10-walkthrough-neuen-demo-mandanten-anlegen)
- [§ 11 — Bekannte Lücken und Trade-offs](#11-bekannte-lücken-und-trade-offs)
## 0. Glossar
| Begriff | Bedeutung |
|---|---|
| **OpenBao** | HashiCorp-Vault-Fork. Single Source of Truth für Secrets. Endpoint: `bao.digitalboard.ch`. |
| **Authentik** | Identity Provider. Stellt OIDC für SP-Services und LDAP via Outpost. |
| **Outpost (Authentik)** | Separater Authentik-Sidecar, der LDAP/Proxy-Protokolle für Legacy-Apps emuliert. Spricht via RPC + Token zu Authentik. |
| **WOPI** | Web Application Open Platform Interface — Protokoll, mit dem Nextcloud/Opencloud Office-Dokumente an Collabora übergeben. |
| **TSIG / RFC2136** | Authenticated DNS-Updates. Traefik nutzt TSIG-signierte `nsupdate`-Calls für ACME DNS-01-Challenges. |
| **DNS-01 (ACME)** | Let's-Encrypt-Challenge-Typ: Zertifikatsbesitz wird per TXT-Record im DNS bewiesen statt per HTTP. Erforderlich für Wildcard-Certs. |
| **CNAME-Bridge** | `_acme-challenge.<fqdn>` zeigt per CNAME in eine dedizierte Update-Zone (`demo-gymb._acme.digitalboard.ch`). So bleibt der TSIG-Key auf eine schmale Zone beschränkt. |
| **File-Provider / Docker-Provider** | Traefik-Konfigurationsquellen. File-Provider liest statische YAML, Docker-Provider liest Container-Labels via `/var/run/docker.sock`. |
| **STUN/TURN** | NAT-Traversal-Protokolle für WebRTC (z. B. für Nextcloud Talk). Läuft auf separatem Host (`turn`). |
| **Garage** | S3-kompatibler Object Store (Rust). Backend für Nextcloud/Opencloud. |
| **FQCN** | Fully Qualified Collection Name, z. B. `digitalboard.core.traefik`. Ansible-Pflicht ab 2.10. |
## 1. Repo-Layout und Roles-Herkunft
```text
reference-ansible/
├── Makefile # Deploy-Targets, OIDC-Login, OBJC-Fork-Workaround
├── ansible.cfg # collections_path, remote_user=root, hashi_vault auth_method=token
├── requirements.yml # community.hashi_vault + digitalboard.core (Git)
├── playbooks/site.yml # Play-Sequenz (14 Plays, siehe § 6)
├── collections/ # ← installiert von `make install`, gitignored
│ └── ansible_collections/
│ └── digitalboard/core/
│ └── roles/ # 🔑 HIER liegen die Rollen, NICHT im Repo-Root
└── inventories/
├── demo-gymburgdorf/ # Inventory dieses Dokuments
├── demo-mbazürich/
├── demo-phbern/
└── vagrant/ # lokale Test-Inventory mit eigener Topologie
```
> **Wichtig:** Es gibt **kein** `roles/`-Verzeichnis im Repo-Root. Alle
> Rollen kommen aus der Collection `digitalboard.core` (siehe
> [requirements.yml](requirements.yml)), installiert via `make install`
> nach `./collections/`. Plays referenzieren sie per FQCN
> `digitalboard.core.<role>`.
## 2. Setup und Voraussetzungen
**Tools auf dem Control-Node:**
- `ansible` (Core ≥ 2.15)
- `bao` CLI (OpenBao) — z. B. `sudo pacman -S openbao python-hvac` (Arch) oder Homebrew
- `python-hvac` (für `community.hashi_vault` Lookups)
- Auf macOS: `OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES` (im [Makefile](Makefile) gesetzt; ohne crashen Ansible-Forks beim Bao-Lookup)
**Initial-Setup:**
```bash
git clone <repo>
cd reference-ansible
make install # Galaxy + digitalboard.core nach ./collections/
```
**Vor jedem Deploy:** Bao-Login in **derselben Shell**, in der dann `ansible-playbook` läuft:
```bash
export BAO_ADDR=https://bao.digitalboard.ch
bao login -method=oidc -path=Digitalboard
export VAULT_TOKEN=$(bao print token)
```
> ⚠️ `make bao` allein reicht **nicht**: jedes `make`-Target startet
> eine neue Shell, der dort gesetzte `VAULT_TOKEN` lebt nur
> während `make bao` selbst. Entweder die drei Befehle oben manuell
> ausführen oder `make bao deploy_site_demo_gymburgdorf` als **einen**
> Aufruf — sonst hat das Deploy keinen Token.
**Smoke-Test:**
```bash
make ping_demo # pingt alle drei Demo-Inventories
```
## 3. Variablen-Hierarchie
Ansible mergt Variablen über mehrere Quellen. Vereinfachtes Modell für
dieses Repo (vollständige Precedence siehe Ansible-Docs):
```mermaid
flowchart LR
classDef role fill:#fef3c7,stroke:#92400e,color:#000
classDef group fill:#dbeafe,stroke:#1e40af,color:#000
classDef host fill:#dcfce7,stroke:#166534,color:#000
classDef vault fill:#fee2e2,stroke:#991b1b,color:#000
R["<b>role defaults</b><br/>(niedrigste Precedence)<br/>collections/.../roles/&lt;r&gt;/defaults/main.yml"]:::role
GA["<b>group_vars/all/</b><br/>vault.yml, docker.yml"]:::group
GG["<b>group_vars/&lt;group&gt;/</b><br/>traefik_servers/, backend_servers/<br/>(parallele Gruppen, gemerged via<br/>ansible_group_priority)"]:::group
HV["<b>host_vars/&lt;host&gt;/</b><br/>(höchste der drei Inventory-Quellen)"]:::host
BAO["<b>OpenBao</b><br/>Lookup zur Laufzeit"]:::vault
R --> |"&lt;wird überschrieben von&gt;"| GA
GA --> |"&lt;wird überschrieben von&gt;"| GG
GG --> |"&lt;wird überschrieben von&gt;"| HV
HV -.community.hashi_vault.-> BAO
GG -.community.hashi_vault.-> BAO
```
**Wichtige Eigenschaften:**
- Mehrere `group_vars/<group>/` sind **parallel**, nicht hierarchisch
geschachtelt. `traefik_servers` und `backend_servers` werden nach
`ansible_group_priority` (Default 1) gemerged; bei Konflikt gewinnt
alphabetisch der spätere Gruppenname.
- `host_vars/<host>/` schlägt jede Gruppe.
- `host_vars/reverseproxy/traefik.yml: traefik_mode: dmz` überschreibt
daher den Default aus `group_vars/backend_servers/` — und zwar nur,
weil `reverseproxy` nicht Mitglied von `backend_servers` ist (sonst
ginge es sowieso nicht).
**Bao-Lookups** sind keine Precedence-Ebene, sondern **Werte** innerhalb
einer beliebigen Var-Quelle. Pattern siehe [§ 8](#8-security-und-demo-only-defaults).
## 4. Inventory-Topologie (`demo-gymburgdorf`)
```mermaid
flowchart LR
classDef dmz fill:#fee2e2,stroke:#991b1b,color:#000
classDef app fill:#dcfce7,stroke:#166534,color:#000
classDef stor fill:#dbeafe,stroke:#1e40af,color:#000
classDef turn fill:#fef9c3,stroke:#854d0e,color:#000
subgraph ALL["group: all_servers"]
direction LR
subgraph DMZ["DMZ 172.16.9.0/24"]
RP["<b>reverseproxy</b><br/>172.16.9.111<br/>traefik_mode: dmz"]:::dmz
TURN["<b>turn</b><br/>172.16.9.112<br/>(noch keine Rolle in site.yml)"]:::turn
end
subgraph BE["Backend 172.16.19.0/24<br/>group: backend_servers"]
APP["<b>application</b><br/>172.16.19.101<br/>traefik_mode: backend<br/>+ authentik, authentik_outpost_ldap,<br/> nextcloud, collabora, drawio"]:::app
ST["<b>storage</b><br/>172.16.19.102<br/>traefik_mode: backend<br/>+ garage (S3)"]:::stor
end
end
RP -.HTTPS in, HTTP out.-> APP
RP -.HTTPS in, HTTP out.-> ST
```
**Gruppen-Mitgliedschaften (aus [hosts.yml](inventories/demo-gymburgdorf/hosts.yml)):**
| Gruppe | Mitglieder | Zweck |
|---|---|---|
| `all_servers` | `reverseproxy`, `application`, `storage`, `turn` | Basis-Rolle für alle Hosts |
| `traefik_servers` | `children: all_servers` (= alle 4 Hosts) | Traefik überall; DMZ/Backend per `traefik_mode` |
| `backend_servers` | `application`, `storage` | setzt `traefik_mode: backend` per group_var |
| `garage_servers` | `storage` | Single-Host-Wrapper für Garage-Role |
| `nextcloud_servers`, `collabora_servers`, `drawio_servers`, `authentik_servers`, `authentik_outpost_ldap_servers` | je nur `application` | Single-Host-Wrapper |
> **Unterschied zur `vagrant`-Inventory:** `vagrant` strukturiert
> Traefik anders — über die Children-Gruppen `traefik_servers_dmz` und
> `traefik_servers_backend` statt über `backend_servers` +
> `host_vars`-Override. Die beiden Topologien sind **strukturell
> inkompatibel**; ein 1:1-Mapping geht nicht. Siehe [§ 10](#10-walkthrough-neuen-demo-mandanten-anlegen)
> für die empfohlene Vorlage.
## 5. Service-Layout und Variablen-Verortung
```mermaid
flowchart TB
classDef rp fill:#fee2e2,stroke:#991b1b,color:#000
classDef ap fill:#dcfce7,stroke:#166534,color:#000
classDef st fill:#dbeafe,stroke:#1e40af,color:#000
classDef ext fill:#e9d5ff,stroke:#6b21a8,color:#000
Internet((Internet))
DNS["DNS ns1.digitalboard.ch<br/>RFC2136 TSIG<br/>Zone: demo-gymb._acme.digitalboard.ch<br/>CNAME-Bridge: _acme-challenge.*.gymb.souveredu.ch"]:::ext
BAO["OpenBao<br/>bao.digitalboard.ch<br/>mount: demo-gymburgdorf"]:::ext
subgraph RP["<b>reverseproxy</b> — traefik dmz"]
TRDMZ["traefik (file provider)<br/>📍 group_vars/traefik_servers/traefik.yml<br/>📍 host_vars/reverseproxy/traefik.yml<br/> → traefik_mode: dmz<br/> → traefik_dmz_exposed_services"]:::rp
end
subgraph APP["<b>application</b> — traefik backend"]
TRA["traefik (docker provider)<br/>📍 group_vars/backend_servers/traefik.yml"]:::ap
AK["authentik (OIDC + LDAP-Outpost-Backend)<br/>📍 host_vars/application/authentik.yml"]:::ap
AKO["authentik_outpost_ldap<br/>📍 host_vars/application/authentik_outpost_ldap.yml"]:::ap
NC["nextcloud<br/>📍 host_vars/application/nextcloud.yml"]:::ap
COL["collabora<br/>📍 host_vars/application/collabora.yml"]:::ap
DRW["drawio<br/>📍 host_vars/application/drawio.yml"]:::ap
end
subgraph ST["<b>storage</b> — traefik backend"]
TRS["traefik (docker provider)"]:::st
GAR["garage (S3)<br/>📍 host_vars/storage/garage.yml"]:::st
end
Internet -->|HTTPS :443| TRDMZ
TRDMZ -->|HTTP backend| TRA
TRDMZ -->|HTTP backend| TRS
TRA --> AK & AKO & NC & COL & DRW
TRS --> GAR
NC -. S3 .-> GAR
NC -. OIDC .-> AK
NC -. WOPI .-> COL
NC -. LDAP .-> AKO
AKO -. RPC + token .-> AK
TRDMZ -. ACME DNS-01 TSIG .-> DNS
TRDMZ -. hashi_vault acme-tsig .-> BAO
AK -. hashi_vault secrets .-> BAO
NC -. hashi_vault secrets .-> BAO
GAR -. hashi_vault secrets .-> BAO
```
> **Hinweis:** `opencloud`, `send` und `openforms` sind in
> [playbooks/site.yml](playbooks/site.yml) als Plays vorhanden, haben
> aber in `demo-gymburgdorf` aktuell keine entsprechende Gruppe in
> [hosts.yml](inventories/demo-gymburgdorf/hosts.yml) — die Plays
> laufen also durch, ohne Targets zu finden. Wenn diese Services für
> einen Mandanten gewünscht sind, jeweils `<service>_servers`-Gruppe in
> `hosts.yml` und `host_vars/application/<service>.yml` ergänzen.
>
> Der Host `turn` ist in `all_servers` und damit auch in
> `traefik_servers`, aber es gibt **keine** Service-Gruppe für ihn —
> aktuell läuft auf `turn` nur die `base`- und `traefik`-Rolle.
## 6. Deploy-Flow
Reihenfolge aus [playbooks/site.yml](playbooks/site.yml):
```mermaid
sequenceDiagram
participant U as User
participant A as ansible-playbook
participant V as OpenBao
participant H as Hosts
U->>U: bao login + export VAULT_TOKEN
U->>A: make deploy_site_demo_gymburgdorf
A->>A: lade vars: role defaults → group_vars/all → group_vars/&lt;groups&gt; → host_vars/&lt;host&gt;
A->>V: community.hashi_vault Lookups<br/>(acme-tsig, service-secrets)
V-->>A: secret values
A->>H: Play 1 — base (alle Hosts)
A->>H: Play 2 — traefik (alle Hosts: dmz auf reverseproxy, backend sonst)
A->>H: Play 3 — httpbin
A->>H: Play 4 — 389ds
A->>H: Play 5 — keycloak
A->>H: Play 6 — garage (storage)
A->>H: Play 7 — collabora (application)
A->>H: Play 8 — authentik (application)
A->>H: Play 9 — authentik_outpost_ldap (application)
A->>H: Play 10 — nextcloud (application)
A->>H: Play 11 — drawio (application)
A->>H: Play 12 — send
A->>H: Play 13 — openforms
A->>H: Play 14 — opencloud
```
Plays ohne passende Gruppen-Mitglieder (`httpbin_servers`, `ds389_servers`,
`keycloak_servers`, `send_servers`, `openforms_servers`,
`opencloud_servers` in dieser Inventory) laufen no-op durch.
`--diff` ist im Target gesetzt → Änderungen pro Task sichtbar.
## 7. Traefik-Modi (DMZ vs Backend)
**`traefik_mode: dmz`** — public-facing Reverse Proxy auf `reverseproxy`:
- **file provider** mit `services.yml` für statisches Routing.
- Kein Docker-Socket gemountet, keine lokalen Container.
- Routet zu `backend_host`-Adressen anderer Maschinen.
- Backends werden über `traefik_dmz_exposed_services` (Liste in
`host_vars/reverseproxy/`) deklariert. Selektive Backend-Auswahl
zusätzlich über `traefik_backend_servers_to_proxy`.
**`traefik_mode: backend`** — application/storage:
- Mountet `/var/run/docker.sock`.
- **docker provider**: Auto-Discovery via Container-Labels (`traefik.enable=true`).
- Services werden lokal exponiert; die DMZ-Traefik routet von aussen
dorthin (Klartext-HTTP, siehe [§ 8](#8-security-und-demo-only-defaults)).
**Beide Modi** unterstützen ACME via RFC2136 DNS Challenge oder
Self-Signed (`traefik_cert_mode: acme | selfsigned`).
## 8. Security und Demo-Only-Defaults
> Dieses Repo ist explizit für **Demo-Setups** gedacht. Alle Default-
> Werte in den Rollen sind unsicher und werden in `demo-*`-Inventories
> über Bao-Lookups oder host_vars überschrieben. Für Prod-Deployments
> gilt zusätzlich der Härtungs-Block weiter unten.
### Secret-Pattern (Bao-Lookup)
```yaml
# group_vars/.../<service>.yml oder host_vars/.../<service>.yml
authentik_secret_key: "{{ lookup('community.hashi_vault.hashi_vault',
vault_mount + '/data/authentik:secret_key',
url=vault_addr) }}"
```
- `vault_mount` und `vault_addr` aus
[group_vars/all/vault.yml](inventories/demo-gymburgdorf/group_vars/all/vault.yml).
- KV-v2-Pfade brauchen explizit `/data/` im Pfad — Ansible löst das
nicht selber auf.
- `vault_mount` ist pro Inventory eindeutig (`demo-gymburgdorf`,
`demo-phbern`, …) → Mandant-Isolation in Bao via Mount + Policy.
### Demo-Only-Defaults — Override-Pflicht
Diese Defaults in `digitalboard.core` sind unsicher. In jeder
**Prod-tauglichen** Deployment müssen sie via Bao-Lookup oder
host_var überschrieben werden:
| Variable | Default | Wo überschreiben |
|---|---|---|
| `keycloak_admin_password` | `changeme` | host_vars `keycloak_servers` |
| `keycloak_postgres_password` | `changeme` | dito |
| `authentik_secret_key` | `changeme-generate-a-random-string` | `host_vars/application/authentik.yml` |
| `authentik_postgres_password` | `changeme` | dito |
| `nextcloud_admin_password` | `admin` | `host_vars/application/nextcloud.yml` |
| `nextcloud_postgres_password` | `changeme` | dito |
| `nextcloud_s3_key` / `nextcloud_s3_secret` | `changeme` / `changeme` | dito |
| `garage_webui_password` | `admin` | `host_vars/storage/garage.yml` |
| `garage_rpc_secret` | `0123…cdef` (64-hex Konstante) | dito |
| `garage_admin_token` | identisch zu `rpc_secret` | dito |
| `garage_metrics_token` | identisch zu `rpc_secret` | dito |
> **Konvention:** Jeder Wert, der oben in der Tabelle steht, **muss**
> in `demo-*/host_vars/.../...yml` einen Bao-Lookup haben, bevor das
> Inventory als deploy-fähig gilt.
### Threat-Boundaries (Stand: Demo)
| Boundary | Status | Notiz |
|---|---|---|
| DMZ ↔ Backend (172.16.9 ↔ 172.16.19) | **Klartext-HTTP** | Auth-Bearer, OIDC-Code, Session-Cookies reisen unverschlüsselt. Für Demo ok, für Prod: mTLS oder WireGuard-Overlay. |
| Host-Firewall | **fehlt** | Die `base`-Rolle installiert kein UFW/nftables. Segmentation hängt am Hypervisor/VLAN. |
| SSH | `ansible_user: root` | Kein Bastion, kein Jumphost. Key-Distribution out-of-band. |
| Authentik-SPOF | **akzeptiert** | IDP und SP-Services auf demselben Host (`application`). Authentik-Ausfall = Login-Ausfall inkl. LDAP-Outpost. Kein Break-Glass-Pfad. |
| ACME-TSIG-Key | Bao-Lookup | Eine TSIG-Key pro Demo-Zone (`acme_update_key_demo_gymb`), Zonen-isoliert. Rotation manuell. |
| Backup/DR | **out-of-scope** | Garage `replication_factor: 1` (default), kein Postgres-Backup-Job, kein Bao-Snapshot-Cron. |
### Für Prod-Adaption ergänzen
- Host-FW (`base`-Rolle erweitern oder eigene `firewall`-Rolle).
- mTLS oder WireGuard zwischen DMZ und Backend.
- Authentik auf separaten Host, mit Recovery-Admin-Token.
- Bao-Policies pro Inventory-Mount (read-only für Deploy-Token,
write-only für Bootstrap-Job).
- Backup-Cron für Postgres + Garage + Bao.
- SSH-Bastion + Key-Rotation.
## 9. Variablen-Cheatsheet
| Variable | Wohin in `demo-gymburgdorf/` | Warum |
|---|---|---|
| `vault_addr`, `vault_mount` | `group_vars/all/vault.yml` | Bao-Endpoint gilt site-weit |
| `docker_registry_mirrors` | `group_vars/all/docker.yml` | Pulls aus Mirror auf allen Hosts |
| `traefik_acme_*`, `traefik_use_ssl`, `traefik_cert_mode` | `group_vars/traefik_servers/traefik.yml` | Gilt für alle Traefik-Instanzen (dmz + backend) |
| `traefik_mode: backend` | `group_vars/backend_servers/traefik.yml` | Default für app + storage |
| `traefik_mode: dmz` | `host_vars/reverseproxy/traefik.yml` | Host-spezifischer Override |
| `traefik_dmz_exposed_services` | `host_vars/reverseproxy/` | Liste der DMZ-Backends — nur dort sinnvoll |
| `nextcloud_*`, `authentik_*`, `collabora_*`, `drawio_*` | `host_vars/application/<service>.yml` | Service läuft auf `application` |
| `garage_*` | `host_vars/storage/garage.yml` | Service läuft auf `storage` |
| Secrets (Passwords, Tokens, Keys) | inline-Variable mit `lookup('community.hashi_vault.hashi_vault', …)` | Single source of truth via Bao |
## 10. Walkthrough: Neuen Demo-Mandanten anlegen
Empfohlene Vorlage: **`demo-gymburgdorf`** (nicht `vagrant`, weil die
Gruppen-Topologie inkompatibel ist).
1. **Inventory kopieren:**
```bash
cp -r inventories/demo-gymburgdorf inventories/demo-<kunde>
```
2. **`hosts.yml` anpassen:** IPs, Hostnames pro Host.
3. **`group_vars/all/vault.yml`** — `vault_mount` auf den neuen
Mandant-Mount setzen (`demo-<kunde>`).
4. **`group_vars/traefik_servers/traefik.yml`** — `traefik_acme_dns_zone`
und `traefik_acme_tsig_*`-Lookup-Pfade auf die neue Zone /
den neuen Bao-Pfad biegen.
5. **`host_vars/application/*.yml`** und **`host_vars/storage/*.yml`**
durchgehen: FQDNs auf das neue Domain-Pattern (z. B.
`*.<kunde>.souveredu.ch`), Bao-Lookup-Pfade auf `demo-<kunde>/data/…`.
6. **OpenBao vorbereiten** (out-of-band, nicht via Ansible):
- Neuen KV-v2-Mount `demo-<kunde>` anlegen.
- Secrets schreiben: `acme-tsig`, `authentik`, `nextcloud`,
`garage`, … (siehe [§ 8](#8-security-und-demo-only-defaults)
für die Override-Pflicht-Liste).
- Policy für den Deploy-Token: read auf `demo-<kunde>/data/*`.
7. **DNS:** TSIG-Update-Zone (`demo-<kunde>._acme.digitalboard.ch`) bei
`ns1.digitalboard.ch` anlegen, CNAMEs `_acme-challenge.*.<kunde>.<tld>`
dorthin.
8. **Makefile** — neues Target nach Vorbild von
`deploy_site_demo_gymburgdorf` ergänzen und in `deploy_site_demo`
einreihen.
9. **Smoke-Test:** `ansible all -i inventories/demo-<kunde>/hosts.yml -m ping`.
10. **Deploy:** Bao-Login + `make deploy_site_demo_<kunde>`.
## 11. Bekannte Lücken und Trade-offs
- **`opencloud`-Inventory in `demo-gymburgdorf`:** keine
`opencloud_servers`-Gruppe vorhanden. Wenn benötigt, ergänzen wie in
[§ 5](#5-service-layout-und-variablen-verortung) beschrieben.
- **`turn`-Host:** in DMZ definiert, aber keine STUN/TURN-Rolle in
[playbooks/site.yml](playbooks/site.yml). Wird aktuell nur per
`base`+`traefik` provisioniert.
- **Idempotenz:** Rollen sind Docker-Compose-basiert; Re-Runs führen
ggf. Container-Restarts aus, wenn sich Compose-Inputs ändern. Kein
spezieller Rollback-Mechanismus — bei Fehlschlag manuell auf den
vorigen Stand zurücksetzen.
- **TLS-Renewal:** Erfolgt durch Traefik intern via ACME. Kein
externer Renew-Cron im Repo.
- **CI/Testing:** Aktuell nicht im Repo. Smoke-Test via `make ping_demo`.
- **Logs:** Traefik läuft in `demo-gymburgdorf` und `vagrant` mit
`traefik_log_level: DEBUG` (Role-Default ist `INFO`) — vor
Prod-Adaption auf `INFO` oder `WARN` reduzieren.

12
Makefile
View file

@ -10,6 +10,18 @@ bao:
bao login -method=oidc -path=Digitalboard role=default bao login -method=oidc -path=Digitalboard role=default
$(eval export VAULT_TOKEN=$(shell bao print token)) $(eval export VAULT_TOKEN=$(shell bao print token))
# Seed/merge OpenBao secrets for a demo inventory. Idempotent: existing
# keys are kept; only missing keys are generated. Pass DRY_RUN=1 to
# preview without writing.
seed_bao_gymburgdorf:
scripts/bao-seed.sh demo-gymburgdorf
seed_bao_mbazürich:
scripts/bao-seed.sh demo-mbazürich
seed_bao_phbern:
scripts/bao-seed.sh demo-phbern
ping_demo: ping_demo:
echo "# pinging demo-gymburgdorf" echo "# pinging demo-gymburgdorf"
ansible all -i inventories/demo-gymburgdorf/hosts.yml -m ping || true ansible all -i inventories/demo-gymburgdorf/hosts.yml -m ping || true

132
README.md
View file

@ -1,61 +1,117 @@
<!-- markdownlint-disable MD013 -->
# reference-ansible # reference-ansible
Ansible-Setup für Demo-Deployments (`demo-gymburgdorf`, ## Purpose
`demo-mbazürich`, `demo-phbern`) und lokale Vagrant-Tests. Rollen
kommen aus der Collection This repository is the **Ansible setup for the Digitalboard demo
deployments** — reproducible reference tenants on which the full stack
(identity, storage, office, forms, dashboards, …) can be set up and
demonstrated end-to-end.
It contains **no** roles of its own: all logic comes from the collection
[`digitalboard.core`](https://git.digitalboard.ch/Digitalboard/digitalboard.core) [`digitalboard.core`](https://git.digitalboard.ch/Digitalboard/digitalboard.core)
(via [requirements.yml](requirements.yml)). (via [requirements.yml](requirements.yml)). What lives here is the
**inventory and configuration layer** — one dedicated inventory per
tenant under `inventories/demo-*` that wires up the roles, defines hosts
and domains, and pulls secrets from [OpenBao](https://openbao.org/). The
only playbook is [playbooks/site.yml](playbooks/site.yml).
> Architektur, Variablen-Hierarchie, Service-Topologie und der The repository thus serves two purposes:
> Walkthrough zum Aufsetzen neuer Mandanten: siehe
> **[ARCHITECTURE.md](ARCHITECTURE.md)**.
## Voraussetzungen - **Demo operation:** deploy and keep up to date the existing demo
tenants (`demo-gymburgdorf`, `demo-mbazürich`, `demo-phbern`).
- **Template:** create new tenants following the `demo-gymburgdorf`
pattern — see [docs/inventories.md § Walkthrough](docs/inventories.md#walkthrough-creating-a-new-demo-tenant).
- `ansible` (Core ≥ 2.15) > **Demo-only.** All role defaults (passwords, tokens, RPC secrets) are
- `bao` CLI ([OpenBao](https://openbao.org/)) — z. B. > insecure and intended exclusively for demo setups. For production
`sudo pacman -S openbao python-hvac` (Arch) oder Homebrew > adaptation, see [docs/secrets.md](docs/secrets.md).
- `python-hvac` (für `community.hashi_vault` Lookups)
## Setup ## Documentation
```bash In-depth documentation lives in the [`docs/`](docs/) folder — start with
make install # installiert digitalboard.core + community.hashi_vault nach ./collections/ the index [docs/README.md](docs/README.md):
| Document | Content |
| --- | --- |
| [docs/getting_started.md](docs/getting_started.md) | Prerequisites (access, tools), first deploy step by step |
| [docs/operations.md](docs/operations.md) | Setup, prerequisites, deploy flow, make targets, smoke test, known gaps |
| [docs/secrets.md](docs/secrets.md) | OpenBao login, secret lookup pattern, demo-only defaults, threat boundaries |
| [docs/inventories.md](docs/inventories.md) | Repository layout, roles origin, inventory topology, new-tenant walkthrough |
| [docs/ansible.md](docs/ansible.md) | Playbooks (`site.yml`), service parameters, variable cheat sheet |
| [docs/testing.md](docs/testing.md) | Static checks, inventory resolution, smoke test/dry run before the deploy |
## Repository structure
```text
reference-ansible/
├── Makefile # deploy targets, OIDC login, OBJC fork workaround
├── ansible.cfg # collections_path, remote_user=root, hashi_vault auth
├── requirements.yml # community.hashi_vault + digitalboard.core (Git)
├── Vagrantfile # local test VMs
├── playbooks/
│ └── site.yml # the only playbook — play sequence of all services
├── scripts/
│ └── bao-seed.sh # seed/merge OpenBao secrets per inventory (idempotent)
├── docs/ # in-depth documentation (see table above)
├── collections/ # ← installed by `make install`, gitignored
│ └── ansible_collections/digitalboard/core/roles/ # 🔑 the roles live here
└── inventories/
├── demo-gymburgdorf/ # reference tenant — template for new tenants
│ ├── hosts.yml # hosts + group topology
│ ├── group_vars/ # all/ · traefik_servers/ · backend_servers/
│ └── host_vars/ # reverseproxy/ · application/ · storage/
├── demo-mbazürich/ # demo tenant
├── demo-phbern/ # demo tenant
└── vagrant/ # local test inventory (incompatible topology)
``` ```
## Secrets (OpenBao) > **No `roles/` in the repository root** — all roles come from
> `digitalboard.core` and are installed via `make install` into
> `./collections/`. Plays reference them by FQCN
> `digitalboard.core.<role>`. Details:
> [docs/inventories.md](docs/inventories.md#repo-layout-and-role-origin).
Vor jedem Deploy in **derselben Shell** authentisieren: ## Quick Start
> **First time here?** Prerequisites (WKS account, OIDC access, SSH key,
> VPN), tool setup, and the first deploy step by step:
> **[docs/getting_started.md](docs/getting_started.md)**.
```bash ```bash
make install # collections into ./collections/
# Log in to OpenBao — in the SAME shell as the deploy.
# Full login flow + make-bao caveat: docs/secrets.md
export BAO_ADDR=https://bao.digitalboard.ch export BAO_ADDR=https://bao.digitalboard.ch
bao login -method=oidc -path=Digitalboard bao login -method=oidc -path=Digitalboard
export VAULT_TOKEN=$(bao print token) export VAULT_TOKEN=$(bao print token)
make ping_demo # smoke test against all demo inventories
make deploy_site_demo_gymburgdorf # single demo site
make deploy_site_demo # all three demo sites
``` ```
> ⚠️ `make bao` allein reicht **nicht** — jedes `make`-Target läuft in Login details and the `make bao` caveat: [docs/secrets.md](docs/secrets.md#openbao-login).
> einer neuen Shell, der dort gesetzte `VAULT_TOKEN` lebt nur während Prerequisites, all make targets, and the deploy flow:
> `make bao` selbst. Entweder die drei Befehle oben manuell im Shell [docs/operations.md](docs/operations.md). Invoking Ansible directly
> ausführen oder `make bao deploy_site_demo_gymburgdorf` als **einen** (`--limit`, `--check`): [docs/ansible.md § Running Ansible](docs/ansible.md#running-ansible).
> Aufruf chainen.
## Deploy ## Available playbooks
```bash The only playbook is [playbooks/site.yml](playbooks/site.yml) — a
make ping_demo # Smoke-Test gegen alle Demo-Inventories sequence of plays, each applying one `digitalboard.core` role to a host
make deploy_site_demo_gymburgdorf # einzelnes Demo-Site group (base, traefik, garage, authentik, authentik_outpost_ldap,
make deploy_site_demo # alle drei Demo-Sites nextcloud, collabora, drawio, send, opnform, homarr, bookstack, …).
``` Which plays take effect in an inventory is governed solely by group
membership in `hosts.yml`. Full play table and all service parameters:
Auf macOS setzt das [Makefile](Makefile) zusätzlich **[docs/ansible.md](docs/ansible.md)**.
`OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES` — ohne diese Env-Var
crashen Ansible-Forks beim ersten `community.hashi_vault`-Lookup.
## Inventories ## Inventories
| Inventory | Zweck | | Inventory | Purpose |
|---|---| | --- | --- |
| [`inventories/demo-gymburgdorf/`](inventories/demo-gymburgdorf/) | Demo-Mandant — als Vorlage für neue Mandanten empfohlen, siehe [ARCHITECTURE.md § 10](ARCHITECTURE.md#10-walkthrough-neuen-demo-mandanten-anlegen) | | [`inventories/demo-gymburgdorf/`](inventories/demo-gymburgdorf/) | Demo tenant — recommended as a template for new tenants, see [docs/inventories.md](docs/inventories.md#walkthrough-creating-a-new-demo-tenant) |
| [`inventories/demo-mbazürich/`](inventories/demo-mbazürich/) | Demo-Mandant | | [`inventories/demo-mbazürich/`](inventories/demo-mbazürich/) | Demo tenant |
| [`inventories/demo-phbern/`](inventories/demo-phbern/) | Demo-Mandant | | [`inventories/demo-phbern/`](inventories/demo-phbern/) | Demo tenant |
| [`inventories/vagrant/`](inventories/vagrant/) | lokale Test-VMs; **inkompatible Gruppen-Topologie** zu den Demo-Inventories | | [`inventories/vagrant/`](inventories/vagrant/) | Local test VMs; **incompatible group topology** compared to the demo inventories |

29
docs/README.md Normal file
View file

@ -0,0 +1,29 @@
<!-- markdownlint-disable MD013 -->
# Documentation — `reference-ansible`
Entry point for this repository's in-depth documentation. The
[`demo-gymburgdorf`](../inventories/demo-gymburgdorf/) inventory serves
as a running example throughout.
> **Demo-only.** All role defaults (passwords, tokens, RPC secrets) are
> insecure and intended exclusively for demo setups. See
> [secrets.md § Demo-Only-Defaults](secrets.md#demo-only-defaults--must-be-overridden).
## Table of contents
| Document | Content |
| --- | --- |
| [getting_started.md](getting_started.md) | Prerequisites (access, tools), first deploy step by step |
| [operations.md](operations.md) | Setup, prerequisites, deploy flow, smoke test, known gaps |
| [secrets.md](secrets.md) | OpenBao login, secret lookup pattern, demo-only defaults, threat boundaries |
| [inventories.md](inventories.md) | Repository layout, roles origin, inventory topology, new-tenant walkthrough |
| [ansible.md](ansible.md) | Playbooks (`site.yml`), per-service parameters, variable cheat sheet |
| [testing.md](testing.md) | Static checks, inventory resolution, smoke test/dry run before the deploy |
## Quick links
- **First time here?** → [getting_started.md](getting_started.md)
- **Create a new tenant** → [inventories.md § Walkthrough](inventories.md#walkthrough-creating-a-new-demo-tenant)
- **Which variable goes where?** → [ansible.md § Variable cheat sheet](ansible.md#variable-cheatsheet)
- **Store a secret in Bao** → [secrets.md § Secret pattern](secrets.md#secret-pattern-bao-lookup)
- **Run a deploy** → [operations.md § Deploy](operations.md#deploy)

209
docs/ansible.md Normal file
View file

@ -0,0 +1,209 @@
<!-- markdownlint-disable MD013 MD060 MD051 -->
# Playbooks & Parameters
[← Documentation index](README.md)
Central reference: which plays [playbooks/site.yml](../playbooks/site.yml)
runs, which service parameters are relevant per role, and where they are
located in the inventory. Example used throughout: `demo-gymburgdorf`.
## Playbook `site.yml`
The only playbook is [playbooks/site.yml](../playbooks/site.yml). It
consists of a sequence of plays, each applying one role from
`digitalboard.core` to a host group. All plays run with
`become: yes`. Plays whose group has no members in an inventory run as a
**no-op**.
| # | Play / role | `hosts:` | Target in `demo-gymburgdorf`? |
| --- | --- | --- | --- |
| 1 | `digitalboard.core.base` | `all_servers` | ✅ all 4 hosts |
| 2 | `digitalboard.core.traefik` | `traefik_servers_backend` | — no-op (vagrant-only group) |
| 3 | `digitalboard.core.traefik` | `traefik_servers_dmz` | — no-op (vagrant-only group) |
| 4 | `digitalboard.core.traefik` | `traefik_servers:!traefik_servers_dmz:!traefik_servers_backend` | ✅ all 4 (dmz on `reverseproxy`, otherwise backend) |
| 5 | `digitalboard.core.httpbin` | `httpbin_servers` | — no-op |
| 6 | `digitalboard.core.389ds` | `ds389_servers` | — no-op |
| 7 | `digitalboard.core.keycloak` | `keycloak_servers` | — no-op |
| 8 | `digitalboard.core.garage` | `garage_servers` | ✅ `storage` |
| 9 | `digitalboard.core.collabora` | `collabora_servers` | ✅ `application` |
| 10 | `digitalboard.core.authentik` | `authentik_servers` | ✅ `application` |
| 11 | `digitalboard.core.authentik_outpost_ldap` | `authentik_outpost_ldap_servers` | ✅ `application` |
| 12 | `digitalboard.core.nextcloud` | `nextcloud_servers` | ✅ `application` |
| 13 | `digitalboard.core.drawio` | `drawio_servers` | ✅ `application` |
| 14 | `digitalboard.core.send` | `send_servers` | ✅ `application` |
| 15 | `digitalboard.core.opnform` | `opnform_servers` | ✅ `application` |
| 16 | `digitalboard.core.homarr` | `homarr_servers` | ✅ `application` |
| 17 | `digitalboard.core.bookstack` | `bookstack_servers` | ✅ `application` |
| 18 | `digitalboard.core.opencloud` | `opencloud_servers` | — no-op (no group) |
> **Three traefik plays, two topologies.** `vagrant` splits the reverse
> proxy into `traefik_servers_dmz` + `traefik_servers_backend` (plays 2
> and 3). The demo inventories (e.g. `demo-gymburgdorf`) instead group
> all hosts under `traefik_servers` and select dmz/backend per host via
> `traefik_mode`; play 4's `:!…` intersection targets exactly those
> hosts and stays a no-op for the vagrant split. Each topology thus
> triggers only the traefik play(s) that fit it — no host runs traefik
> twice.
>
> Which plays take effect for a tenant is controlled **solely through
> group membership** in `hosts.yml`. A service becomes active as soon as
> its `<service>_servers` group contains a host and a matching
> `host_vars/<host>/<service>.yml` exists.
## Running Ansible
**Prerequisite:** collections installed (`make install`) and logged in
to OpenBao in the **same shell** (`VAULT_TOKEN` set) — without a token,
the `community.hashi_vault` lookups fail. Login procedure:
[secrets.md § OpenBao login](secrets.md#openbao-login).
Initial setup step by step: [getting_started.md](getting_started.md).
### Via Makefile (recommended)
```bash
make ping_demo # Smoke test (ping) against all demo inventories
make deploy_site_demo_gymburgdorf # single demo site
make deploy_site_demo # all three demo sites in sequence
```
The Make targets encapsulate the full `ansible-playbook` invocation
including `--diff` and the macOS fork env var. All targets:
[operations.md § Makefile reference](operations.md#makefile-reference).
### Direct `ansible-playbook` invocation
When you need flags that the targets do not set:
```bash
# Full deploy of an inventory
ansible-playbook playbooks/site.yml \
-i inventories/demo-gymburgdorf/hosts.yml --diff
# Only one host (e.g. just the application machine)
ansible-playbook playbooks/site.yml \
-i inventories/demo-gymburgdorf/hosts.yml --limit application
# Dry run without changes
ansible-playbook playbooks/site.yml \
-i inventories/demo-gymburgdorf/hosts.yml --check --diff
```
> Because service selection runs through the groups in `hosts.yml` (not
> through tags), `--limit <host>` is the usual way to narrow down a
> deploy. `--check` is only of limited value with the Docker Compose-based
> roles — some tasks report "changed" because they only learn the actual
> container state at runtime.
Deploy flow and play order: [operations.md § Deploy](operations.md#deploy).
## Where parameters belong
| Variable group | File in `demo-gymburgdorf/` | Why |
| --- | --- | --- |
| `vault_addr`, `vault_mount` | `group_vars/all/vault.yml` | Bao endpoint applies site-wide |
| `docker_registry_mirrors` | `group_vars/all/docker.yml` | Pulls from mirror on all hosts |
| `traefik_acme_*`, `traefik_use_ssl`, `traefik_cert_mode`, `traefik_log_level` | `group_vars/traefik_servers/traefik.yml` | applies to all Traefik instances (dmz + backend) |
| `traefik_mode: backend` | `group_vars/backend_servers/traefik.yml` | default for app + storage |
| `traefik_mode: dmz`, `traefik_dmz_exposed_services` | `host_vars/reverseproxy/traefik.yml` | host-specific override, only meaningful there |
| `nextcloud_*`, `authentik_*`, `collabora_*`, `drawio_*`, `send_*`, `opnform_*`, `homarr_*`, `bookstack_*` | `host_vars/application/<service>.yml` | service runs on `application` |
| `garage_*` | `host_vars/storage/garage.yml` | service runs on `storage` |
| Secrets (passwords, tokens, keys) | inline var with `lookup('community.hashi_vault.hashi_vault', …)` | single source of truth via Bao, see [secrets.md](secrets.md) |
## Service parameters in detail
Complete variable lists are in the `defaults/main.yml` of the respective
role in `digitalboard.core`. Below are the parameters maintained in the
demo inventories per service — as guidance on which fields a new tenant
typically needs to set.
### traefik
| Variable | Example / default | Purpose |
| --- | --- | --- |
| `traefik_mode` | `dmz` \| `backend` | Provider mode: `dmz` = file provider (public-facing, no Docker socket), `backend` = docker provider (auto-discovery via container labels) |
| `traefik_cert_mode` | `acme` \| `selfsigned` | Certificate source |
| `traefik_use_ssl` | `true` | TLS active |
| `traefik_ssl_email` | `hostmaster@digitalboard.ch` | ACME contact |
| `traefik_log_level` | `DEBUG` (role default `INFO`) | reduce for prod |
| `traefik_network` | `proxy` | Docker network for backend mode |
| `traefik_acme_dns_zone` | `demo-gymb._acme.digitalboard.ch` | RFC2136 update zone |
| `traefik_acme_dns_nameserver` | from Bao / `172.16.9.169` (DMZ override) | TSIG update target |
| `traefik_acme_tsig_algorithm` / `_key` / `_secret` | `hmac-sha256` / Bao | TSIG signature |
| `traefik_acme_tcp_only` | `true` | force DNS lookups over TCP/53 |
| `traefik_acme_disable_ans_checks` | `true` (DMZ only) | skip NS propagation poll |
| `traefik_dmz_exposed_services` | list (DMZ) | which backends the DMZ Traefik routes |
### authentik (IdP — OIDC + LDAP outpost backend)
| Variable | Purpose |
| --- | --- |
| `authentik_domains` | public FQDNs (`auth.gymb.souveredu.ch`) |
| `authentik_host_rewrite_domains` | internal `*.int.*` names for LAN server-to-server |
| `authentik_secret_key`, `authentik_postgres_password` | Bao lookup |
| `authentik_ldap_apps`, `authentik_ldap_outpost` | LDAP app + outpost definition (base_dn, token) |
| `authentik_proxy_apps` | ForwardAuth apps (slug, external/internal_host, allowed_groups) |
### authentik_outpost_ldap
| Variable | Purpose |
| --- | --- |
| `authentik_outpost_ldap_host` | internal Authentik host (`https://auth.int.…`) |
| `authentik_outpost_ldap_token` | outpost token (Bao, identical to `authentik.ldap_outpost_token`) |
### nextcloud
| Variable | Purpose |
| --- | --- |
| `nextcloud_image` | image tag (pin to patched version) |
| `nextcloud_domains` | first entry = canonical public FQDN, further `*.int.*` |
| `nextcloud_admin_user` / `_password`, `nextcloud_postgres_password` | admin + DB (Bao) |
| `nextcloud_use_s3_storage`, `nextcloud_s3_*` | S3 primary storage via Garage (key/secret via `garage_credentials` lookup) |
| `nextcloud_enable_collabora`, `nextcloud_collabora_domain` / `_public_domain` | WOPI integration |
| `nextcloud_enable_drawio`, `nextcloud_drawio_url` | Draw.io integration |
| `nextcloud_oidc_providers` | OIDC login via Authentik (discovery_url, client_id/secret) |
| `nextcloud_ldap_enabled`, `nextcloud_ldap_config` | LDAP backend against Authentik outpost |
| `nextcloud_apps_to_install` | app list (groupfolders, richdocuments, spreed, user_ldap, …) |
| `nextcloud_allow_local_remote_servers`, `nextcloud_extra_hosts` | LAN-only routing for server-to-server calls |
### collabora
| Variable | Purpose |
| --- | --- |
| `collabora_domains` | public + internal FQDN |
| `collabora_allowed_domains`, `collabora_frame_ancestors` | allowed WOPI hosts / iframe embedding |
### drawio
| Variable | Purpose |
| --- | --- |
| `drawio_domain`, `drawio_extra_domains` | public + internal FQDN |
| `drawio_authentik_forward_auth`, `_url` | access protection via Authentik ForwardAuth |
### garage (S3 object store)
| Variable | Purpose |
| --- | --- |
| `garage_s3_domains` | first entry = public S3 FQDN, further `*.int.*` |
| `garage_webui_domain`, `garage_webui_enabled` | admin WebUI |
| `garage_webui_authentik_forward_auth`, `_url` | WebUI behind Authentik (admins only) |
| `garage_rpc_secret`, `garage_admin_token`, `garage_metrics_token` | Bao lookup |
| `garage_bootstrap_*` | single-node cluster bootstrap (zone, capacity) |
| `garage_s3_keys` | keys + buckets + permissions (e.g. `nextcloud`) |
### send / opnform / homarr / bookstack
Same pattern: `<service>_domain`/`_domains` (+ `*.int.*`),
`<service>_base_url`, admin credentials and app keys via Bao lookup,
plus OIDC integration with Authentik (`<service>_oidc_*`: issuer, client_id,
client_secret, admin_group). For the concrete fields, see the respective
`host_vars/application/<service>.yml`.
## Variable cheatsheet
Short form of the location table above — "which variable goes where":
- **Site-wide** → `group_vars/all/` (Bao endpoint, Docker mirror)
- **All Traefik** → `group_vars/traefik_servers/`
- **app + storage** → `group_vars/backend_servers/`
- **Single host** → `host_vars/<host>/<service>.yml`
- **Secrets** → always Bao lookup, never plaintext (see [secrets.md](secrets.md))

93
docs/getting_started.md Normal file
View file

@ -0,0 +1,93 @@
<!-- markdownlint-disable MD013 -->
# Getting Started
[← Documentation index](README.md)
From zero to your first deploy. This page walks through prerequisites,
setup, and the first Ansible run. Deeper details are linked along the way.
## Prerequisites
### Access (to be set up out-of-band)
- **WKS account with OIDC access to OpenBao.** The login runs via
`bao login -method=oidc -path=Digitalboard`. Without an authorized
account, authentication fails — and without a token there is no
secret lookup, hence no deploy.
- **Bao policy / mount read.** The account needs **Read** on the
mount of the target inventory (e.g. `demo-gymburgdorf/data/*`). Which
paths an inventory reads is documented in the `host_vars/.../<service>.yml`
(see [secrets.md § Secret pattern](secrets.md#secret-pattern-bao-lookup)).
- **SSH key on the target hosts.** The hosts are provisioned as `root`
(`ansible_user: root`, no bastion/jump host). Your own public key
must be placed out-of-band as `root` on the hosts.
- **Network access (VPN).** `bao.digitalboard.ch` and the host networks
(`172.16.9.0/24` DMZ, `172.16.19.0/24` backend) are not publicly
reachable — access requires VPN/network access into the Digitalboard network.
### Tools on the control node
- `ansible` (Core ≥ 2.15) — `ansible --version` to check
- `bao` CLI ([OpenBao](https://openbao.org/)) — e.g.
`sudo pacman -S openbao python-hvac` (Arch) or via Homebrew
- `python-hvac` (for `community.hashi_vault` lookups)
- On macOS: `OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES` (set in the
[Makefile](../Makefile)) — without this env var, Ansible forks crash
on the first Bao lookup, because the Objective-C runtime is not
fork-safe.
## 1. Clone the repo and install collections
```bash
git clone https://git.digitalboard.ch/Digitalboard/reference-ansible
cd reference-ansible
make install # community.hashi_vault + digitalboard.core into ./collections/
```
There is **no** `roles/` directory in the repo — all roles come from
the `digitalboard.core` collection. See
[inventories.md § Repo layout](inventories.md#repo-layout-and-role-origin).
## 2. Log in to OpenBao
The login must happen in the **same shell** in which
`ansible-playbook` then runs — details and the `make bao` caveat in
[secrets.md § OpenBao login](secrets.md#openbao-login):
```bash
export BAO_ADDR=https://bao.digitalboard.ch
bao login -method=oidc -path=Digitalboard
export VAULT_TOKEN=$(bao print token)
```
## 3. Check connectivity (smoke test)
```bash
make ping_demo # ping module against all three demo inventories
```
If a host does not respond, it is usually due to the SSH key
(prerequisite above) or missing network access (VPN).
## 4. Run Ansible (deploy)
```bash
make deploy_site_demo_gymburgdorf # single demo site
```
At its core, `make` only calls `ansible-playbook` — the equivalent
direct invocation:
```bash
ansible-playbook playbooks/site.yml \
-i inventories/demo-gymburgdorf/hosts.yml --diff
```
All variants (direct invocation, `--limit`, `--tags`, check mode) and
the make targets are documented in [ansible.md § Running Ansible](ansible.md#running-ansible).
## Next steps
- **What happens during a deploy?** → [ansible.md § Playbook](ansible.md#playbook-siteyml)
- **Create a new tenant** → [inventories.md § Walkthrough](inventories.md#walkthrough-creating-a-new-demo-tenant)
- **Store secrets in Bao** → [secrets.md](secrets.md)

179
docs/inventories.md Normal file
View file

@ -0,0 +1,179 @@
<!-- markdownlint-disable MD013 MD060 MD051 -->
# Repo layout & inventories
[← Documentation index](README.md)
## Repo layout and role origin
```text
reference-ansible/
├── Makefile # deploy targets, OIDC login, OBJC fork workaround
├── ansible.cfg # collections_path, remote_user=root, hashi_vault auth_method=token
├── requirements.yml # community.hashi_vault + digitalboard.core (Git)
├── playbooks/site.yml # play sequence (see ansible.md)
├── scripts/bao-seed.sh # seed/merge OpenBao secrets per inventory
├── docs/ # this documentation
├── collections/ # ← installed by `make install`, gitignored
│ └── ansible_collections/
│ └── digitalboard/core/
│ └── roles/ # 🔑 THE ROLES LIVE HERE, NOT in the repo root
└── inventories/
├── demo-gymburgdorf/ # reference inventory of this documentation
├── demo-mbazürich/
├── demo-phbern/
└── vagrant/ # local test inventory with its own topology
```
> **Important:** There is **no** `roles/` directory in the repo root. All
> roles come from the `digitalboard.core` collection (see
> [requirements.yml](../requirements.yml)), installed via
> `make install` into `./collections/`. Plays reference them by
> FQCN `digitalboard.core.<role>`.
## Available inventories
| Inventory | Purpose |
| --- | --- |
| [`demo-gymburgdorf/`](../inventories/demo-gymburgdorf/) | Demo tenant — **recommended as the template for new tenants** |
| [`demo-mbazürich/`](../inventories/demo-mbazürich/) | Demo tenant |
| [`demo-phbern/`](../inventories/demo-phbern/) | Demo tenant |
| [`vagrant/`](../inventories/vagrant/) | local test VMs; **incompatible group topology** with the demo inventories |
## Inventory topology (`demo-gymburgdorf`)
```mermaid
flowchart LR
classDef dmz fill:#fee2e2,stroke:#991b1b,color:#000
classDef app fill:#dcfce7,stroke:#166534,color:#000
classDef stor fill:#dbeafe,stroke:#1e40af,color:#000
classDef turn fill:#fef9c3,stroke:#854d0e,color:#000
subgraph ALL["group: all_servers"]
direction LR
subgraph DMZ["DMZ 172.16.9.0/24"]
RP["<b>reverseproxy</b><br/>172.16.9.111<br/>traefik_mode: dmz"]:::dmz
TURN["<b>turn</b><br/>172.16.9.112<br/>(no role in site.yml yet)"]:::turn
end
subgraph BE["Backend 172.16.19.0/24<br/>group: backend_servers"]
APP["<b>application</b><br/>172.16.19.101<br/>traefik_mode: backend<br/>+ authentik, authentik_outpost_ldap,<br/> nextcloud, collabora, drawio, …"]:::app
ST["<b>storage</b><br/>172.16.19.102<br/>traefik_mode: backend<br/>+ garage (S3)"]:::stor
end
end
RP -.HTTPS in, HTTP out.-> APP
RP -.HTTPS in, HTTP out.-> ST
```
**Group memberships (from [hosts.yml](../inventories/demo-gymburgdorf/hosts.yml)):**
| Group | Members | Purpose |
| --- | --- | --- |
| `all_servers` | `reverseproxy`, `application`, `storage`, `turn` | base role for all hosts |
| `traefik_servers` | `children: all_servers` (all 4 hosts) | Traefik everywhere; DMZ/backend via `traefik_mode` |
| `backend_servers` | `application`, `storage` | sets `traefik_mode: backend` via group_var |
| `garage_servers` | `storage` | single-host wrapper for the Garage role |
| `nextcloud_servers`, `collabora_servers`, `drawio_servers`, `authentik_servers`, `authentik_outpost_ldap_servers`, `send_servers`, `opnform_servers`, `homarr_servers`, `bookstack_servers` | only `application` each | single-host wrappers |
> **Difference from the `vagrant` inventory:** `vagrant` structures
> Traefik differently — via the children groups `traefik_servers_dmz` and
> `traefik_servers_backend` instead of via `backend_servers` +
> `host_vars` override. The two topologies are **structurally
> incompatible**; a 1:1 mapping is not possible. For new tenants, therefore,
> take `demo-gymburgdorf` as the template.
## Standard folder structure of an inventory entry
A fully built-out inventory follows this layout (example
`demo-gymburgdorf`). Currently only this inventory is built out;
`demo-mbazürich` and `demo-phbern` so far contain only `hosts.yml`.
```text
inventories/demo-<kunde>/
├── hosts.yml # REQUIRED — hosts, IPs, group topology
├── group_vars/
│ ├── all/
│ │ ├── vault.yml # REQUIRED — vault_addr, vault_mount (Bao)
│ │ ├── ansible.yml # ansible_python_interpreter etc.
│ │ └── docker.yml # docker_registry_mirrors
│ ├── traefik_servers/
│ │ └── traefik.yml # ACME/TSIG, TLS — applies to ALL Traefik instances
│ └── backend_servers/
│ └── traefik.yml # traefik_mode: backend (default for app + storage)
└── host_vars/
├── reverseproxy/
│ └── traefik.yml # traefik_mode: dmz + DMZ-specific ACME overrides
├── application/
│ ├── main.yml # comment only: which services run here
│ ├── traefik.yml # traefik_dmz_exposed_services (what the DMZ routes)
│ └── <service>.yml # one file per service (nextcloud, authentik, …)
└── storage/
├── main.yml # same as above
├── traefik.yml # traefik_extra_hosts + traefik_dmz_exposed_services
└── garage.yml # service vars for garage
```
**Conventions:**
- **`hosts.yml` is the only hard required file.** Vars are
optional — if one is missing, the role defaults from
`digitalboard.core` take effect. A new inventory therefore starts minimally with
only `hosts.yml` (just like `demo-mbazürich`/`demo-phbern`).
- **`group_vars/all/vault.yml`** is effectively required as soon as
Bao lookups are supposed to work — without `vault_mount`/`vault_addr` the
secret lookups fail.
- **One file per service** under `host_vars/<host>/<service>.yml`. The
file name is free (Ansible loads all YAMLs in the directory); by
convention it is named like the role. Which variables belong where:
[ansible.md § Where parameters belong](ansible.md#where-parameters-belong).
- **`main.yml` per host** is pure documentation — a comment indicating which
services run on the host. Carries no productive vars.
- **`host_vars/<host>/traefik.yml`** declares via
`traefik_dmz_exposed_services` which local services the
DMZ Traefik should make reachable from outside. The DMZ reads this
list via `hostvars[<backend>]` and renders its routers from it. A new
service exposed externally = a new entry here. Mechanics:
[ansible.md § traefik](ansible.md#traefik).
## Walkthrough: Creating a new demo tenant
Recommended template: **`demo-gymburgdorf`** (not `vagrant`, because its
group topology is incompatible).
1. **Copy the inventory:**
```bash
cp -r inventories/demo-gymburgdorf inventories/demo-<kunde>
```
2. **Adjust `hosts.yml`:** IPs, hostnames per host.
3. **`group_vars/all/vault.yml`** — set `vault_mount` to the new
tenant mount (`demo-<kunde>`).
4. **`group_vars/traefik_servers/traefik.yml`** —
point `traefik_acme_dns_zone` and the `acme-tsig` lookup paths to the
new zone / the new Bao path.
5. Go through **`host_vars/application/*.yml`** and **`host_vars/storage/*.yml`**:
FQDNs to the new domain pattern (e.g.
`*.<kunde>.souveredu.ch`), Bao lookup paths to `demo-<kunde>/data/…`.
6. **Prepare OpenBao** (out-of-band, not via Ansible):
- Create a new KV-v2 mount `demo-<kunde>`.
- Write secrets: `acme-tsig`, `authentik`, `nextcloud`,
`garage`, … — conveniently via `make seed_bao_<kunde>` (see
[scripts/bao-seed.sh](../scripts/bao-seed.sh) and
[secrets.md § Demo-Only-Defaults](secrets.md#demo-only-defaults--must-be-overridden)).
- Policy for the deploy token: read on `demo-<kunde>/data/*`.
7. **DNS:** Create the TSIG update zone (`demo-<kunde>._acme.digitalboard.ch`) at
`ns1.digitalboard.ch`, CNAMEs
`_acme-challenge.*.<kunde>.<tld>` pointing there.
8. **Makefile** — add a new target modeled on
`deploy_site_demo_gymburgdorf` and add it to `deploy_site_demo`;
likewise a `seed_bao_<kunde>` target.
9. **Smoke test:** `ansible all -i inventories/demo-<kunde>/hosts.yml -m ping`.
10. **Deploy:** Bao login + `make deploy_site_demo_<kunde>`.

136
docs/operations.md Normal file
View file

@ -0,0 +1,136 @@
<!-- markdownlint-disable MD013 -->
# Setup & operations
[← Documentation index](README.md)
## Prerequisites (control node)
- `ansible` (Core ≥ 2.15)
- `bao` CLI ([OpenBao](https://openbao.org/)) — e.g.
`sudo pacman -S openbao python-hvac` (Arch) or via Homebrew
- `python-hvac` (for `community.hashi_vault` lookups)
- On macOS: `OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES` (set in the
[Makefile](../Makefile); without this env var, Ansible forks crash
on the first `community.hashi_vault` lookup, because the
Objective-C runtime is not fork-safe)
## Initial setup
```bash
git clone <repo>
cd reference-ansible
make install # Galaxy + digitalboard.core into ./collections/
```
`make install` installs `community.hashi_vault` and the
`digitalboard.core` collection (Git, see [requirements.yml](../requirements.yml))
into `./collections/`. There is **no** `roles/` directory in the
repo root — all roles come from the collection, see
[inventories.md § Repo layout](inventories.md#repo-layout-and-role-origin).
## Secrets (OpenBao)
Before every deploy, authenticate to OpenBao in the **same shell**. The
full login flow, the `make bao` caveat, the lookup pattern, and
tenant isolation are documented in **[secrets.md](secrets.md)**.
## Smoke test
```bash
make ping_demo # pings all three demo inventories (ping module)
```
## Deploy
```bash
make deploy_site_demo_gymburgdorf # single demo site
make deploy_site_demo_mbazürich
make deploy_site_demo_phbern
make deploy_site_demo # all three in sequence
```
`--diff` is set in the Gymburgdorf target → changes are visible per task.
The play order and which plays run as no-ops:
see [ansible.md § Playbooks](ansible.md#playbook-siteyml).
## Makefile reference
The [Makefile](../Makefile) bundles setup, secret handling, and deploy.
It defines no variables for passing in except `DRY_RUN` (for the
`seed_bao_*` targets) — control is via the chosen target.
### Exported env vars (apply to all targets)
| Variable | Value | Purpose |
| --- | --- | --- |
| `BAO_ADDR` | `https://bao.digitalboard.ch` | OpenBao endpoint for `bao` and `community.hashi_vault` calls |
| `OBJC_DISABLE_INITIALIZE_FORK_SAFETY` | `YES` | macOS fork safety: without this var, Ansible forks crash on the first `hashi_vault` lookup, because the Objective-C runtime is not fork-safe |
> Both are set via `export` at the top of the Makefile and thus
> inherited by every target shell process — regardless of which target runs.
### Setup & secrets
| Target | Effect |
| --- | --- |
| `make install` | `ansible-galaxy collection install -r requirements.yml -p collections` — installs `community.hashi_vault` + `digitalboard.core` into `./collections/` |
| `make bao` | `bao login -method=oidc -path=Digitalboard role=default` + sets `VAULT_TOKEN` via `$(eval …)`. ⚠️ The token only lives **within this single `make` invocation** — see caveat below |
| `make seed_bao_gymburgdorf` | Seed/merge OpenBao secrets for `demo-gymburgdorf` via [scripts/bao-seed.sh](../scripts/bao-seed.sh). Idempotent: existing keys remain, only missing ones are generated |
| `make seed_bao_mbazürich` | same for `demo-mbazürich` |
| `make seed_bao_phbern` | same for `demo-phbern` |
> The `seed_bao_*` targets understand `DRY_RUN=1` — shows the diff without
> writing: `make seed_bao_gymburgdorf DRY_RUN=1`. Requirement:
> `bao`, `jq`, `openssl` in `$PATH` and a valid `VAULT_TOKEN`.
### Smoke test & deploy
| Target | Effect |
| --- | --- |
| `make ping_demo` | `ansible … -m ping` against all three demo inventories in sequence; failures of individual hosts do not abort (`\|\| true`) |
| `make deploy_site_demo_gymburgdorf` | `ansible-playbook playbooks/site.yml -i …/demo-gymburgdorf/hosts.yml --diff` |
| `make deploy_site_demo_mbazürich` | same for `demo-mbazürich` — **without** `--diff` |
| `make deploy_site_demo_phbern` | same for `demo-phbern` — **without** `--diff` |
| `make deploy_site_demo` | calls the three `deploy_site_demo_*` targets in sequence |
> **Inconsistency:** Only the Gymburgdorf target sets `--diff`. For
> `mbazürich` and `phbern` you do not see the task changes — if
> needed, invoke directly with `ansible-playbook … --diff`, see
> [ansible.md § Running Ansible](ansible.md#running-ansible).
### Token caveat (`make bao`)
`make bao` alone is **not** enough for a deploy: each `make` target
runs in its own shell, the `VAULT_TOKEN` set there only lives
during `make bao` itself and is already gone in the next `make deploy_…`.
Two working approaches:
```bash
# Variant A — log in manually in the active shell (survives multiple make invocations)
export BAO_ADDR=https://bao.digitalboard.ch
bao login -method=oidc -path=Digitalboard
export VAULT_TOKEN=$(bao print token)
make deploy_site_demo_gymburgdorf
# Variant B — chain both as ONE make invocation (token lives for the chain)
make bao deploy_site_demo_gymburgdorf
```
Login details and the secret pattern: [secrets.md](secrets.md#openbao-login).
## Known gaps and trade-offs
- **`opencloud` in `demo-gymburgdorf`:** Play present, but no
`opencloud_servers` group — runs as a no-op. If needed, add a group +
`host_vars`, see [inventories.md](inventories.md#walkthrough-creating-a-new-demo-tenant).
- **`turn` host:** defined in the DMZ, but no STUN/TURN role in
[playbooks/site.yml](../playbooks/site.yml) — provisioned only via `base` +
`traefik`.
- **Idempotency:** Roles are Docker-Compose-based; re-runs can
trigger container restarts when Compose inputs change. No
rollback mechanism — on failure, roll back manually.
- **TLS renewal:** handled internally by Traefik via ACME, no external
renew cron in the repo.
- **CI/testing:** currently not in the repo; smoke test via `make ping_demo`.
- **Logging:** `traefik_log_level: DEBUG` in `demo-gymburgdorf` and
`vagrant` (role default `INFO`) — reduce before adapting to prod.

99
docs/secrets.md Normal file
View file

@ -0,0 +1,99 @@
<!-- markdownlint-disable MD013 MD060 MD051 -->
# Secrets & security
[← Documentation index](README.md)
> This repo is explicitly intended for **demo setups**. All
> default values in the roles are insecure and are overridden in
> `demo-*` inventories via Bao lookups or host_vars.
## OpenBao login
A prerequisite is a WKS account with OIDC access to OpenBao and a
read policy on the inventory mount — see
[getting_started.md § Vorbedingungen](getting_started.md#prerequisites).
Before each deploy, authenticate in **the same shell** in which
`ansible-playbook` then runs:
```bash
export BAO_ADDR=https://bao.digitalboard.ch
bao login -method=oidc -path=Digitalboard
export VAULT_TOKEN=$(bao print token)
```
> ⚠️ `make bao` alone is **not** enough — every `make` target runs in
> a new shell, and the `VAULT_TOKEN` set there lives only during
> `make bao` itself. Either run the three commands above manually
> or chain `make bao deploy_site_demo_gymburgdorf` as **one** call
> — otherwise the deploy has no token.
## Secret pattern (Bao lookup)
Secrets are never stored in plaintext, but read from
OpenBao at runtime:
```yaml
# host_vars/.../<service>.yml — one lookup per service path,
# individual keys as properties:
_nextcloud: "{{ lookup('community.hashi_vault.hashi_vault',
vault_mount + '/data/nextcloud', url=vault_addr) }}"
nextcloud_admin_password: "{{ _nextcloud.admin_password }}"
nextcloud_postgres_password: "{{ _nextcloud.postgres_password }}"
```
- `vault_mount` and `vault_addr` come from
[group_vars/all/vault.yml](../inventories/demo-gymburgdorf/group_vars/all/vault.yml).
- KV-v2 paths need an explicit `/data/` in the path — Ansible does not
resolve this on its own.
- `vault_mount` is unique per inventory (`demo-gymburgdorf`,
`demo-phbern`, …) → tenant isolation in Bao via mount + policy.
Secrets are seeded idempotently with [scripts/bao-seed.sh](../scripts/bao-seed.sh) (or
`make seed_bao_<kunde>`): existing keys remain,
only missing ones are generated. OIDC client secrets are kept in sync between
`<mount>/data/authentik` and the respective service secret.
## Demo-only defaults — must be overridden
These defaults in `digitalboard.core` are insecure. In every
**production-grade** deployment they must be overridden via a Bao lookup or host_var:
| Variable | Default | Where to override |
| --- | --- | --- |
| `keycloak_admin_password` | `changeme` | host_vars `keycloak_servers` |
| `keycloak_postgres_password` | `changeme` | same as above |
| `authentik_secret_key` | `changeme-generate-a-random-string` | `host_vars/application/authentik.yml` |
| `authentik_postgres_password` | `changeme` | same as above |
| `nextcloud_admin_password` | `admin` | `host_vars/application/nextcloud.yml` |
| `nextcloud_postgres_password` | `changeme` | same as above |
| `nextcloud_s3_key` / `nextcloud_s3_secret` | `changeme` / `changeme` | same as above |
| `garage_webui_password` | `admin` | `host_vars/storage/garage.yml` |
| `garage_rpc_secret` | `0123…cdef` (64-hex constant) | same as above |
| `garage_admin_token` | identical to `rpc_secret` | same as above |
| `garage_metrics_token` | identical to `rpc_secret` | same as above |
> **Convention:** Every value above **must** have a Bao lookup in
> `demo-*/host_vars/.../...yml` before the
> inventory counts as deployable.
## Threat boundaries (status: demo)
| Boundary | Status | Note |
| --- | --- | --- |
| DMZ ↔ backend (172.16.9 ↔ 172.16.19) | **plaintext HTTP** | auth bearer, OIDC code, session cookies travel unencrypted. Demo-ok; prod: mTLS or WireGuard overlay. |
| Host firewall | **missing** | The `base` role installs no UFW/nftables. Segmentation depends on the hypervisor/VLAN. |
| SSH | `ansible_user: root` | No bastion, no jump host. Key distribution out-of-band. |
| Authentik SPOF | **accepted** | IdP and SP services on the same host (`application`). Authentik outage = login outage including LDAP outpost. No break-glass path. |
| ACME TSIG key | Bao lookup | One TSIG key per demo zone (`acme_update_key_demo_gymb`), zone-isolated. Rotation manual. |
| Backup/DR | **out-of-scope** | Garage `replication_factor: 1`, no Postgres backup job, no Bao snapshot cron. |
## Add for production adaptation
- Host FW (extend the `base` role or a dedicated `firewall` role).
- mTLS or WireGuard between DMZ and backend.
- Authentik on a separate host, with a recovery admin token.
- Bao policies per inventory mount (read-only for the deploy token,
write-only for the bootstrap job).
- Backup cron for Postgres + Garage + Bao.
- SSH bastion + key rotation.

93
docs/testing.md Normal file
View file

@ -0,0 +1,93 @@
<!-- markdownlint-disable MD013 -->
# Testing
[← Documentation index](README.md)
> **Status:** This repo contains **no** automated test suite
> and **no** CI pipeline. It is the inventory/configuration
> layer — the testable logic (roles) lives in
> [`digitalboard.core`](https://git.digitalboard.ch/Digitalboard/digitalboard.core).
> Role tests (Molecule or similar) therefore belong in the core repo, not
> here.
What is sensibly testable here are **inventory and playbook errors
before the actual deploy**. There are three levels for that — from fast and
risk-free to fully against the hosts.
## 1. Static checks (no host access needed)
No `VAULT_TOKEN`, no network access required:
```bash
# YAML syntax of all inventory files
yamllint inventories/
# Ansible best-practice lint over the playbook
ansible-lint playbooks/site.yml
# Playbook syntax + inventory parsing (does not yet resolve Bao lookups)
ansible-playbook playbooks/site.yml \
-i inventories/demo-gymburgdorf/hosts.yml --syntax-check
```
`yamllint` and `ansible-lint` are not preconfigured in the repo —
they run with their defaults. If project-wide rules are desired,
a `.yamllint`/`.ansible-lint` in the repo root would be the place for it
(see [Open items](#open-items)).
## 2. Inspect inventory resolution
Shows the effectively merged variables per host — useful for seeing
precedence surprises (group_vars vs. host_vars) before the deploy.
Bao lookups are evaluated here, so `VAULT_TOKEN` is needed (see
[secrets.md](secrets.md#openbao-login)):
```bash
# group/host structure as a tree
ansible-inventory -i inventories/demo-gymburgdorf/hosts.yml --graph
# all vars of a host (merged)
ansible-inventory -i inventories/demo-gymburgdorf/hosts.yml --host application
```
## 3. Smoke test & dry run (against the hosts)
Requires SSH access and `VAULT_TOKEN` — for prerequisites see
[getting_started.md](getting_started.md#prerequisites):
```bash
# reachability of all demo hosts (ping module)
make ping_demo
# dry run: shows what WOULD change, without writing
ansible-playbook playbooks/site.yml \
-i inventories/demo-gymburgdorf/hosts.yml --check --diff
```
> `--check` is only of limited value with the Docker-Compose-based roles:
> some tasks report "changed" because they only know the real
> container state at runtime. As a plausibility check
> (does the playbook run through, are the vars correct?) it is still
> useful. More on the invocation variants:
> [ansible.md § Running Ansible](ansible.md#running-ansible).
## Recommended pre-deploy workflow
```bash
yamllint inventories/ && ansible-lint playbooks/site.yml # 1. static
ansible-playbook playbooks/site.yml -i <inv>/hosts.yml --syntax-check
make ping_demo # 2. reachability
ansible-playbook playbooks/site.yml -i <inv>/hosts.yml --check --diff # 3. dry run
# only then: real deploy
```
## Open items
- **No CI** — lint/syntax checks run only manually. A Gitea/
CI workflow that runs the static checks from level 1 on every push
would be the next step.
- **No lint configuration** — `yamllint`/`ansible-lint` run with
defaults; project-wide rules (`.yamllint`, `.ansible-lint`) are missing.
- **Role tests** — belong in
[`digitalboard.core`](https://git.digitalboard.ch/Digitalboard/digitalboard.core),
not in this repo.

96
inventories/demo-gymburgdorf/host_vars/application/authentik.yml
View file

@ -2,15 +2,21 @@
# Bao secret expected at <mount>/data/authentik with keys: # Bao secret expected at <mount>/data/authentik with keys:
# secret_key, postgres_password, admin_password, # secret_key, postgres_password, admin_password,
# ldap_outpost_token, # ldap_outpost_token,
# nextcloud_oidc_secret # nextcloud_oidc_secret,
# opnform_oidc_secret, homarr_oidc_secret, bookstack_oidc_secret
_authentik: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/authentik', url=vault_addr) }}" _authentik: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/authentik', url=vault_addr) }}"
# First entry is the canonical public FQDN. Additional entries cover # Canonical public FQDN browsers and OIDC iss-claim use.
# internal *.int.* names so server-to-server traffic (e.g. the LDAP
# outpost) hits authentik on a name with a valid internal cert and
# skips the DMZ hop.
authentik_domains: authentik_domains:
- "auth.gymb.souveredu.ch" - "auth.gymb.souveredu.ch"
# Internal FQDN for server-to-server calls (Nextcloud OIDC discovery,
# token, userinfo; LDAP outpost configuration pull). Traefik rewrites
# the Host header to `authentik_domains[0]` on these routers so authentik
# still emits issuer URLs against the public hostname — that keeps the
# iss claim matching what the browser sees while the traffic itself
# stays inside the LAN (the DMZ has no hairpin-NAT for the public IP).
authentik_host_rewrite_domains:
- "auth.int.gymb.souveredu.ch" - "auth.int.gymb.souveredu.ch"
authentik_secret_key: "{{ _authentik.secret_key }}" authentik_secret_key: "{{ _authentik.secret_key }}"
authentik_postgres_password: "{{ _authentik.postgres_password }}" authentik_postgres_password: "{{ _authentik.postgres_password }}"
@ -31,6 +37,44 @@ authentik_ldap_outpost:
authentik_host: "https://auth.int.gymb.souveredu.ch/" authentik_host: "https://auth.int.gymb.souveredu.ch/"
log_level: "info" log_level: "info"
# Proxy providers (ForwardAuth) — gate downstream services behind
# authentik. The embedded outpost (which authentik ships out of the box)
# hosts these providers under /outpost.goauthentik.io/auth/traefik on the
# canonical FQDN; the service-side traefik attaches a ForwardAuth
# middleware that talks to that endpoint.
authentik_proxy_apps:
- slug: drawio
name: Drawio
external_host: "https://draw.gymb.souveredu.ch"
internal_host: "http://drawio:8080"
allowed_groups:
- admins
flows:
authentication_slug: default-authentication-flow
authorization_slug: default-provider-authorization-implicit-consent
invalidation_slug: default-provider-invalidation-flow
- slug: garage-webui
name: "Garage S3 Console"
external_host: "https://console.s3.gymb.souveredu.ch"
internal_host: "http://garage-webui:3909"
allowed_groups:
- admins
flows:
authentication_slug: default-authentication-flow
authorization_slug: default-provider-authorization-implicit-consent
invalidation_slug: default-provider-invalidation-flow
# Bind both proxy providers to authentik's built-in embedded outpost so
# we don't have to deploy a separate proxy outpost container. The
# embedded outpost listens on the same host:9000 as the authentik server
# and exposes /outpost.goauthentik.io/auth/traefik for ForwardAuth.
authentik_proxy_outposts:
- name: "authentik Embedded Outpost"
type: proxy
providers:
- drawio
- garage-webui
# OIDC clients # OIDC clients
authentik_oidc_apps: authentik_oidc_apps:
- slug: nextcloud - slug: nextcloud
@ -45,10 +89,52 @@ authentik_oidc_apps:
authorization_slug: default-provider-authorization-implicit-consent authorization_slug: default-provider-authorization-implicit-consent
invalidation_slug: default-provider-invalidation-flow invalidation_slug: default-provider-invalidation-flow
scopes: [openid, email, profile, offline_access] scopes: [openid, email, profile, offline_access]
- slug: opnform
name: OpnForm
client_id: opnform
client_secret: "{{ _authentik.opnform_oidc_secret }}"
redirect_uris:
- url: "https://forms.gymb.souveredu.ch/auth/authentik/callback"
matching_mode: strict
signing_key_name: "authentik Self-signed Certificate"
flows:
authorization_slug: default-provider-authorization-implicit-consent
invalidation_slug: default-provider-invalidation-flow
# No separate `groups` scope — authentik's default `profile` mapping
# already emits a `groups` claim built from request.user.groups, so
# OpnForm's admin-group mapping works without an extra scope.
scopes: [openid, email, profile]
- slug: homarr
name: Homarr
client_id: homarr
client_secret: "{{ _authentik.homarr_oidc_secret }}"
redirect_uris:
- url: "https://home.gymb.souveredu.ch/api/auth/callback/oidc"
matching_mode: strict
signing_key_name: "authentik Self-signed Certificate"
flows:
authorization_slug: default-provider-authorization-implicit-consent
invalidation_slug: default-provider-invalidation-flow
scopes: [openid, email, profile]
- slug: bookstack
name: BookStack
client_id: bookstack
client_secret: "{{ _authentik.bookstack_oidc_secret }}"
redirect_uris:
- url: "https://wiki.gymb.souveredu.ch/oidc/callback"
matching_mode: strict
signing_key_name: "authentik Self-signed Certificate"
flows:
authorization_slug: default-provider-authorization-implicit-consent
invalidation_slug: default-provider-invalidation-flow
scopes: [openid, email, profile]
authentik_groups: authentik_groups:
- name: admins - name: admins
- name: users - name: users
- name: opnform-admins
- name: homarr-admins
- name: bookstack-admins
authentik_local_users: authentik_local_users:
- username: akadmin - username: akadmin

43
inventories/demo-gymburgdorf/host_vars/application/bookstack.yml Normal file
View file

@ -0,0 +1,43 @@
---
# Bao secret <mount>/data/bookstack expected to contain:
# db_root_password, db_password, admin_password, oidc_client_secret,
# app_key (optional — only set when restoring)
_bookstack: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/bookstack', url=vault_addr) }}"
bookstack_domain: "wiki.gymb.souveredu.ch"
bookstack_extra_domains:
- "wiki.int.gymb.souveredu.ch"
bookstack_base_url: "https://wiki.gymb.souveredu.ch"
# Override the role-default certresolver ("le") with the value used
# across this demo (matches traefik_ssl_cert_resolver in group_vars).
bookstack_traefik_certresolver: "dns"
bookstack_db_root_password: "{{ _bookstack.db_root_password }}"
bookstack_db_password: "{{ _bookstack.db_password }}"
bookstack_admin_password: "{{ _bookstack.admin_password }}"
bookstack_admin_email: "admin@gymb.souveredu.ch"
bookstack_admin_name: "BookStack Admin"
# OIDC against Authentik. BookStack compares OIDC_ISSUER strictly against
# the `iss` claim in the discovery response. Authentik emits the public
# auth.gymb.* hostname there (host-rewrite middleware ensures the claim
# matches what browsers see during login), so the issuer URL must use the
# public FQDN. Pinning auth.gymb.* in /etc/hosts below keeps the actual
# server-to-server traffic on the LAN.
bookstack_oidc_enabled: true
bookstack_oidc_name: "Authentik"
bookstack_oidc_issuer: "https://auth.gymb.souveredu.ch/application/o/bookstack/"
bookstack_oidc_client_id: "bookstack"
bookstack_oidc_client_secret: "{{ _bookstack.oidc_client_secret }}"
bookstack_oidc_additional_scopes: "openid profile email"
bookstack_oidc_user_to_groups: true
bookstack_oidc_groups_claim: "groups"
bookstack_oidc_auto_initiate: false
# Pin auth.gymb.* to the application host so server-to-server OIDC calls
# (discovery, token, userinfo, jwks) stay in the LAN and reach authentik
# directly without hairpinning through the DMZ (which has no NAT loop
# back to its own public IP).
bookstack_extra_hosts:
- "auth.gymb.souveredu.ch:172.16.19.101"

17
inventories/demo-gymburgdorf/host_vars/application/drawio.yml
View file

@ -1,2 +1,19 @@
--- ---
drawio_domain: "draw.gymb.souveredu.ch" drawio_domain: "draw.gymb.souveredu.ch"
# Internal FQDN the DMZ reverseproxy uses as backend host so its TLS
# verify matches a cert SAN (the canonical IP-only route has no SAN
# and breaks with "cannot validate certificate ... no IP SANs"). Same
# split-horizon pattern as cloud.int.* / auth.int.* / office.int.*.
drawio_extra_domains:
- "draw.int.gymb.souveredu.ch"
# Gate drawio behind the authentik embedded outpost (admins-only —
# enforced by the policy-binding on the authentik proxy application).
# ForwardAuth talks to the embedded outpost on the authentik server's
# in-network address. Going via the public FQDN routes through a second
# traefik hop that strips/rewrites X-Forwarded-Host, which breaks
# authentik's provider matching (it returns 404). Plain HTTP to the
# container is the path docs recommend for the embedded outpost.
drawio_authentik_forward_auth: true
drawio_authentik_forward_auth_url: "http://authentik-server-1:9000/outpost.goauthentik.io/auth/traefik"

71
inventories/demo-gymburgdorf/host_vars/application/homarr.yml Normal file
View file

@ -0,0 +1,71 @@
---
# Bao secret <mount>/data/homarr expected to contain:
# secret_encryption_key (64 hex chars), admin_password, oidc_client_secret
_homarr: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/homarr', url=vault_addr) }}"
homarr_domain: "home.gymb.souveredu.ch"
homarr_extra_domains:
- "home.int.gymb.souveredu.ch"
homarr_base_url: "https://home.gymb.souveredu.ch"
homarr_secret_encryption_key: "{{ _homarr.secret_encryption_key }}"
homarr_admin_username: "admin"
homarr_admin_email: "admin@gymb.souveredu.ch"
homarr_admin_password: "{{ _homarr.admin_password }}"
# OIDC against Authentik. credentials provider stays enabled as a
# break-glass account.
homarr_auth_providers: "credentials,oidc"
homarr_oidc_issuer: "https://auth.int.gymb.souveredu.ch/application/o/homarr/"
homarr_oidc_client_id: "homarr"
homarr_oidc_client_secret: "{{ _homarr.oidc_client_secret }}"
homarr_oidc_client_name: "Authentik"
homarr_oidc_scopes: "openid profile email groups"
homarr_oidc_groups_attribute: "groups"
homarr_oidc_admin_group: "homarr-admins"
homarr_oidc_auto_login: "false"
# Default board with shortcuts to the other gymburgdorf services. Width
# values describe horizontal grid cells (1-10 desktop / 6 tablet / 2
# mobile, packed left-to-right).
homarr_apps:
- id: nextcloud
name: Nextcloud
description: "Cloud Storage & Collaboration"
icon: https://cdn.jsdelivr.net/gh/walkxcode/dashboard-icons/png/nextcloud.png
href: https://cloud.gymb.souveredu.ch
width: 2
- id: collabora
name: Collabora Office
icon: https://cdn.jsdelivr.net/gh/walkxcode/dashboard-icons/png/collaboraonline.png
href: https://office.gymb.souveredu.ch
width: 2
- id: drawio
name: Draw.io
icon: https://cdn.jsdelivr.net/gh/walkxcode/dashboard-icons/png/drawio.png
href: https://draw.gymb.souveredu.ch
width: 2
- id: send
name: Send
description: "Encrypted file-share"
icon: https://cdn.jsdelivr.net/gh/walkxcode/dashboard-icons/png/firefox-send.png
href: https://send.gymb.souveredu.ch
width: 2
- id: opnform
name: OpnForm
description: "Self-hosted forms"
icon: https://cdn.jsdelivr.net/gh/walkxcode/dashboard-icons/png/opnform.png
href: https://forms.gymb.souveredu.ch
width: 2
- id: bookstack
name: BookStack
description: "Wiki & documentation"
icon: https://cdn.jsdelivr.net/gh/walkxcode/dashboard-icons/png/bookstack.png
href: https://wiki.gymb.souveredu.ch
width: 2
- id: authentik
name: Authentik
description: "Identity provider"
icon: https://cdn.jsdelivr.net/gh/walkxcode/dashboard-icons/png/authentik.png
href: https://auth.gymb.souveredu.ch
width: 2

33
inventories/demo-gymburgdorf/host_vars/application/nextcloud.yml
View file

@ -4,6 +4,11 @@
_nextcloud: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/nextcloud', url=vault_addr) }}" _nextcloud: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/nextcloud', url=vault_addr) }}"
_authentik: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/authentik', url=vault_addr) }}" _authentik: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/authentik', url=vault_addr) }}"
# 33.0.2 hits the PHP 8.4 TypeError in UserConfig::getValueBool() that
# user_ldap triggers on every authenticated request (nextcloud/server
# #59629; fix in 33.0.3). Pin to the patched tag.
nextcloud_image: "nextcloud:33.0.3-fpm"
# First entry is the canonical public FQDN (used for OVERWRITEHOST and # First entry is the canonical public FQDN (used for OVERWRITEHOST and
# OIDC redirects). Additional entries cover internal *.int.* names so # OIDC redirects). Additional entries cover internal *.int.* names so
# collabora's WOPI callbacks hit nextcloud on a name with a valid # collabora's WOPI callbacks hit nextcloud on a name with a valid
@ -57,10 +62,25 @@ nextcloud_s3_port: 443
nextcloud_s3_ssl: true nextcloud_s3_ssl: true
nextcloud_s3_usepath_style: true nextcloud_s3_usepath_style: true
# OIDC server-to-server discovery / token / userinfo goes to
# auth.int.gymb.souveredu.ch (LAN, RFC1918). Nextcloud's DnsPinMiddleware
# would otherwise block that as "local server access".
nextcloud_allow_local_remote_servers: true
# Share the LDAP docker network with the authentik LDAP outpost # Share the LDAP docker network with the authentik LDAP outpost
nextcloud_extra_networks: nextcloud_extra_networks:
- ldap - ldap
# Pin the public authentik FQDN to the application host so server-to-server
# OIDC traffic (token, userinfo, jwks — endpoints the discovery doc lists
# under auth.gymb.* even when discovery itself is fetched via auth.int.*)
# stays in the LAN. Without this, curl in the PHP container would hit the
# public IP and time out in the DMZ (no hairpin-NAT). The DnsPin middleware
# only honours /etc/hosts when allow_local_remote_servers is enabled, so
# that flag (set above) is what makes this entry effective.
nextcloud_extra_hosts:
- "auth.gymb.souveredu.ch:172.16.19.101"
# LDAP backend (Authentik LDAP outpost) # LDAP backend (Authentik LDAP outpost)
nextcloud_ldap_enabled: true nextcloud_ldap_enabled: true
nextcloud_ldap_config: nextcloud_ldap_config:
@ -98,12 +118,13 @@ nextcloud_oidc_providers:
display_name: "Login with Authentik" display_name: "Login with Authentik"
client_id: nextcloud client_id: nextcloud
client_secret: "{{ _authentik.nextcloud_oidc_secret }}" client_secret: "{{ _authentik.nextcloud_oidc_secret }}"
# Stays on the public FQDN: user_oidc validates the iss claim against # Discovery via the internal FQDN (LAN-only) — the DMZ has no
# the discovery host, and authentik returns iss based on the request # hairpin-NAT for the public IP, so server-to-server calls to
# host — using auth.int.* would break the iss match with what the # auth.gymb.* would time out. The traefik router for auth.int.*
# browser sees (auth.gymb.*). Routed via the DMZ for now; revisit if # rewrites the Host header to auth.gymb.souveredu.ch before the
# this becomes a bottleneck. # request reaches authentik, so the iss claim authentik emits still
discovery_url: "https://auth.gymb.souveredu.ch/application/o/nextcloud/.well-known/openid-configuration" # matches the public hostname the browser sees during login.
discovery_url: "https://auth.int.gymb.souveredu.ch/application/o/nextcloud/.well-known/openid-configuration"
scope: "openid email profile" scope: "openid email profile"
unique_uid: true unique_uid: true
mapping: mapping:

60
inventories/demo-gymburgdorf/host_vars/application/opnform.yml Normal file
View file

@ -0,0 +1,60 @@
---
# Bao secret <mount>/data/opnform expected to contain:
# app_key (must start with "base64:"), jwt_secret, front_api_secret,
# db_password, admin_password, oidc_client_secret
_opnform: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/opnform', url=vault_addr) }}"
_authentik: "{{ lookup('community.hashi_vault.hashi_vault', vault_mount + '/data/authentik', url=vault_addr) }}"
opnform_domain: "forms.gymb.souveredu.ch"
opnform_extra_domains:
- "forms.int.gymb.souveredu.ch"
opnform_base_url: "https://forms.gymb.souveredu.ch"
opnform_app_key: "{{ _opnform.app_key }}"
opnform_jwt_secret: "{{ _opnform.jwt_secret }}"
opnform_front_api_secret: "{{ _opnform.front_api_secret }}"
opnform_db_password: "{{ _opnform.db_password }}"
# Bootstrap admin via API on first run so the manual setup page is
# skipped. The admin credentials are also required to seed the OIDC
# IdentityConnection through OpnForm's API (only an authenticated admin
# can create connections).
opnform_admin_name: "OpnForm Admin"
opnform_admin_email: "admin@gymb.souveredu.ch"
opnform_admin_password: "{{ _opnform.admin_password }}"
# OIDC against Authentik. Discovery via the internal FQDN keeps
# server-to-server traffic in the LAN; Authentik's host-rewrite router
# rewrites the Host header to auth.gymb.* before the request reaches
# authentik so the iss claim still matches the public hostname browsers
# see during login.
opnform_oidc_enabled: true
# Issuer must use the public auth.gymb.* FQDN: OpnForm does OIDC
# discovery and then validates the token's `iss` claim against this
# value. Authentik emits the public hostname in `iss` (its host-rewrite
# middleware keeps the claim aligned with what browsers see), so an
# internal-FQDN issuer here would fail iss validation. The extra_hosts
# pin below keeps the actual discovery/token/userinfo traffic on the LAN.
opnform_oidc_issuer: "https://auth.gymb.souveredu.ch/application/o/opnform/"
opnform_oidc_client_id: "opnform"
opnform_oidc_client_secret: "{{ _opnform.oidc_client_secret }}"
opnform_oidc_client_name: "Authentik"
opnform_oidc_slug: "authentik"
opnform_oidc_domain: "gymb.souveredu.ch"
opnform_oidc_admin_group: "opnform-admins"
# Disable password login entirely — every user goes through Authentik.
# All real users have @gymb.souveredu.ch addresses (matching
# opnform_oidc_domain above), so no password fallback is needed.
opnform_oidc_force_login: true
# Serve a /sso page that jumps straight to Authentik without the email
# login form. Link users to https://forms.gymb.souveredu.ch/sso.
opnform_oidc_sso_entrypoint: true
# Pin auth.gymb.* to the application host so server-to-server OIDC
# calls (token, userinfo, jwks — endpoints discovery returns under the
# public hostname even when discovery itself is fetched via auth.int.*)
# stay in the LAN.
opnform_extra_hosts:
- "auth.gymb.souveredu.ch:172.16.19.101"

8
inventories/demo-gymburgdorf/host_vars/application/send.yml Normal file
View file

@ -0,0 +1,8 @@
---
# Send: anonymized self-hosted file-share (no login). First entry is the
# canonical public FQDN (used as BASE_URL); the *.int.* entry covers the
# server-to-server hop from the DMZ reverseproxy with a cert SAN that
# matches the backend hostname (same split-horizon pattern as cloud/draw).
send_domains:
- "send.gymb.souveredu.ch"
- "send.int.gymb.souveredu.ch"

25
inventories/demo-gymburgdorf/host_vars/application/traefik.yml
View file

@ -21,9 +21,26 @@ traefik_dmz_exposed_services:
protocol: https protocol: https
- name: drawio - name: drawio
domain: draw.gymb.souveredu.ch domain: draw.gymb.souveredu.ch
# No internal FQDN/cert for drawio yet — proxy by IP. Combined backend_host: draw.int.gymb.souveredu.ch
# with serversTransport `insecureSkipVerify` (handled by the port: 443
# selfsigned-mode branch in the template), or accept the route's protocol: https
# 500 until the cert is wired up. - name: send
domain: send.gymb.souveredu.ch
backend_host: send.int.gymb.souveredu.ch
port: 443
protocol: https
- name: opnform
domain: forms.gymb.souveredu.ch
backend_host: forms.int.gymb.souveredu.ch
port: 443
protocol: https
- name: homarr
domain: home.gymb.souveredu.ch
backend_host: home.int.gymb.souveredu.ch
port: 443
protocol: https
- name: bookstack
domain: wiki.gymb.souveredu.ch
backend_host: wiki.int.gymb.souveredu.ch
port: 443 port: 443
protocol: https protocol: https

12
inventories/demo-gymburgdorf/host_vars/storage/garage.yml
View file

@ -12,8 +12,18 @@ garage_s3_domains:
garage_webui_domain: "console.s3.gymb.souveredu.ch" garage_webui_domain: "console.s3.gymb.souveredu.ch"
garage_use_ssl: true garage_use_ssl: true
garage_webui_enabled: true garage_webui_enabled: true
# Gate the WebUI behind authentik (admins-only, via policy-binding on the
# authentik proxy app). Replaces the htpasswd Basic-Auth — AUTH_USER_PASS
# is dropped from the compose env when this is true. The forwardauth URL
# resolves to the application-host traefik (network alias
# `auth.gymb.souveredu.ch` -> authentik-server-1 in the proxy network on
# the application host), but THIS host (storage) is in a different LAN,
# so traefik here reaches it via the public name through the DMZ proxy.
garage_webui_authentik_forward_auth: true
garage_webui_authentik_forward_auth_url: "https://auth.gymb.souveredu.ch/outpost.goauthentik.io/auth/traefik"
# Kept for completeness — only used when authentik ForwardAuth is off.
garage_webui_username: "admin" garage_webui_username: "admin"
garage_webui_password: "{{ _garage.webui_password }}" garage_webui_password: "{{ _garage.webui_password | default('disabled') }}"
garage_rpc_secret: "{{ _garage.rpc_secret }}" garage_rpc_secret: "{{ _garage.rpc_secret }}"
garage_admin_token: "{{ _garage.admin_token }}" garage_admin_token: "{{ _garage.admin_token }}"

8
inventories/demo-gymburgdorf/host_vars/storage/traefik.yml
View file

@ -1,4 +1,12 @@
--- ---
# Local traefik needs to reach authentik for the ForwardAuth subrequest
# the garage-webui router fires. The public IP is unreachable from this
# subnet (no DMZ hairpin), so point auth.gymb.* directly at the
# application host where authentik runs. Without this the forwardauth
# middleware would time out and every garage-console request would 502.
traefik_extra_hosts:
- "auth.gymb.souveredu.ch:172.16.19.101"
# Services hosted on `storage` that the DMZ reverseproxy should forward # Services hosted on `storage` that the DMZ reverseproxy should forward
# public traffic to. See application/traefik.yml for the mechanism. # public traffic to. See application/traefik.yml for the mechanism.
traefik_dmz_exposed_services: traefik_dmz_exposed_services:

16
inventories/demo-gymburgdorf/hosts.yml
View file

@ -47,3 +47,19 @@ all:
authentik_outpost_ldap_servers: authentik_outpost_ldap_servers:
hosts: hosts:
application: application:
send_servers:
hosts:
application:
opnform_servers:
hosts:
application:
homarr_servers:
hosts:
application:
bookstack_servers:
hosts:
application:

47
playbooks/site.yml
View file

@ -17,6 +17,17 @@
roles: roles:
- digitalboard.core.traefik - digitalboard.core.traefik
# Inventories without the _dmz/_backend split (e.g. demo-gymburgdorf,
# where traefik_servers groups all_servers and dmz/backend is selected
# per host via traefik_mode). The :!… intersection keeps this a no-op
# for the vagrant topology, where every traefik_servers host is already
# covered by the two plays above.
- name: Configure reverse proxies
hosts: traefik_servers:!traefik_servers_dmz:!traefik_servers_backend
become: yes
roles:
- digitalboard.core.traefik
- name: Deploy httpbin service - name: Deploy httpbin service
hosts: httpbin_servers hosts: httpbin_servers
become: yes become: yes
@ -71,23 +82,17 @@
roles: roles:
- digitalboard.core.drawio - digitalboard.core.drawio
# - name: Deploy send service - name: Deploy send service
# hosts: send_servers hosts: send_servers
# become: yes
# roles:
# - digitalboard.core.send
# - name: Deploy openforms service
# hosts: openforms_servers
# become: yes
# roles:
# - digitalboard.core.openforms
- name: Deploy opencloud service
hosts: opencloud_servers
become: yes become: yes
roles: roles:
- digitalboard.core.opencloud - digitalboard.core.send
- name: Deploy opnform service
hosts: opnform_servers
become: yes
roles:
- digitalboard.core.opnform
- name: Deploy homarr service - name: Deploy homarr service
hosts: homarr_servers hosts: homarr_servers
@ -95,8 +100,14 @@
roles: roles:
- digitalboard.core.homarr - digitalboard.core.homarr
- name: Deploy opnform service - name: Deploy bookstack service
hosts: opnform_servers hosts: bookstack_servers
become: yes become: yes
roles: roles:
- digitalboard.core.opnform - digitalboard.core.bookstack
- name: Deploy opencloud service
hosts: opencloud_servers
become: yes
roles:
- digitalboard.core.opencloud

191
scripts/bao-seed.sh Executable file
View file

@ -0,0 +1,191 @@
#!/usr/bin/env bash
#
# Seed OpenBao secrets for a demo inventory. Merge semantics: existing
# keys at a given path are kept; only missing keys are generated. OIDC
# client secrets are synced between <mount>/data/authentik and the
# per-service secret so authentik and the service agree on the same
# value.
#
# Usage:
# scripts/bao-seed.sh demo-gymburgdorf
#
# Requirements:
# - bao CLI in $PATH, authenticated (make bao)
# - jq in $PATH
# - openssl in $PATH
#
# Environment overrides:
# BAO_ADDR default https://bao.digitalboard.ch
# DRY_RUN=1 print what would change without writing
set -eu
MOUNT="${1:-}"
if [[ -z "$MOUNT" ]]; then
echo "usage: $0 <bao-mount> (e.g. demo-gymburgdorf)" >&2
exit 1
fi
: "${BAO_ADDR:=https://bao.digitalboard.ch}"
export BAO_ADDR
for cmd in bao jq openssl mktemp; do
command -v "$cmd" >/dev/null || { echo "missing: $cmd" >&2; exit 1; }
done
if ! bao token lookup >/dev/null 2>&1; then
echo "not authenticated — run 'make bao' first" >&2
exit 1
fi
DRY_RUN="${DRY_RUN:-0}"
WORKDIR="$(mktemp -d)"
trap 'rm -rf "$WORKDIR"' EXIT
# Read current data for a KV-v2 secret into $WORKDIR/<path>.json.
# Writes {} if the secret does not exist or has no data yet.
read_secret() {
local path="$1"
local out="$WORKDIR/$path.json"
local raw="$WORKDIR/$path.raw"
if bao kv get -format=json "$MOUNT/$path" >"$raw" 2>/dev/null; then
if jq -e -c '.data.data // {}' "$raw" >"$out" 2>/dev/null; then
return 0
fi
fi
echo '{}' >"$out"
}
# Write the in-memory secret back to bao using `bao kv put @file` so
# values containing `=` (e.g. base64 padding) survive intact.
write_secret() {
local path="$1"
local src="$WORKDIR/$path.json"
if [[ "$DRY_RUN" == "1" ]]; then
echo " [dry-run] would write $MOUNT/$path"
return
fi
bao kv put "$MOUNT/$path" "@$src" >/dev/null
}
# Print the current value of a key from a secret file, empty string if
# absent.
get_key() {
local file="$1"
local key="$2"
jq -r --arg k "$key" '.[$k] // ""' "$file"
}
# Set a key inside the secret file to the given literal value.
set_key() {
local file="$1"
local key="$2"
local value="$3"
local tmp="$file.tmp"
jq --arg k "$key" --arg v "$value" '.[$k] = $v' "$file" >"$tmp"
mv "$tmp" "$file"
}
# Generate a key in the secret file if missing.
ensure_key() {
local path="$1"
local key="$2"
local generator="$3"
local file="$WORKDIR/$path.json"
if [[ -n "$(get_key "$file" "$key")" ]]; then
return
fi
local value
value="$($generator)"
set_key "$file" "$key" "$value"
echo " + $key"
}
# Force a key on path2 to match path1's value for the named key.
# Prints when a change happens.
sync_key_from() {
local src_path="$1"
local src_key="$2"
local dst_path="$3"
local dst_key="$4"
local src_file="$WORKDIR/$src_path.json"
local dst_file="$WORKDIR/$dst_path.json"
local src_val
src_val="$(get_key "$src_file" "$src_key")"
if [[ -z "$src_val" ]]; then
echo " ! $src_path/$src_key missing — cannot sync to $dst_path/$dst_key" >&2
return
fi
local dst_val
dst_val="$(get_key "$dst_file" "$dst_key")"
if [[ "$src_val" == "$dst_val" ]]; then
return
fi
set_key "$dst_file" "$dst_key" "$src_val"
echo " = $dst_key (synced from $src_path/$src_key)"
}
gen_hex32() { openssl rand -hex 32; }
gen_hex64() { openssl rand -hex 64; }
gen_pass() { openssl rand -base64 24 | tr -d '/+='; }
gen_long_pass() { openssl rand -base64 32 | tr -d '/+='; }
gen_app_key() { echo "base64:$(openssl rand -base64 32)"; }
# OpnForm requires: min 8 chars, a letter, a digit, AND one of the
# special chars @$!%*#?&-_+=.,:;<>^()[]{}|~. The base64 generator alone
# only produces [A-Za-z0-9], so we append a fixed-position special char
# and digit to guarantee the rule passes regardless of entropy outcome.
gen_opnform_pass() { echo "$(openssl rand -base64 18 | tr -d '/+=')!1Aa"; }
# ---------------------------------------------------------------- main
echo "==> seeding $MOUNT (dry-run=$DRY_RUN)"
echo "-> authentik"
read_secret authentik
ensure_key authentik secret_key gen_hex64
ensure_key authentik postgres_password gen_pass
ensure_key authentik admin_password gen_pass
ensure_key authentik ldap_outpost_token gen_hex32
ensure_key authentik nextcloud_oidc_secret gen_hex32
ensure_key authentik opnform_oidc_secret gen_hex32
ensure_key authentik homarr_oidc_secret gen_hex32
ensure_key authentik bookstack_oidc_secret gen_hex32
write_secret authentik
echo "-> opnform"
read_secret opnform
ensure_key opnform app_key gen_app_key
ensure_key opnform jwt_secret gen_hex32
ensure_key opnform front_api_secret gen_hex32
ensure_key opnform db_password gen_long_pass
ensure_key opnform admin_password gen_opnform_pass
ensure_key opnform oidc_client_secret gen_hex32
sync_key_from authentik opnform_oidc_secret opnform oidc_client_secret
write_secret opnform
echo "-> homarr"
read_secret homarr
ensure_key homarr secret_encryption_key gen_hex32
ensure_key homarr admin_password gen_pass
ensure_key homarr oidc_client_secret gen_hex32
sync_key_from authentik homarr_oidc_secret homarr oidc_client_secret
write_secret homarr
echo "-> bookstack"
read_secret bookstack
ensure_key bookstack db_root_password gen_long_pass
ensure_key bookstack db_password gen_long_pass
ensure_key bookstack admin_password gen_pass
ensure_key bookstack oidc_client_secret gen_hex32
sync_key_from authentik bookstack_oidc_secret bookstack oidc_client_secret
write_secret bookstack
echo "==> done"