M6: OpenWebUI intern produktiv + Produktentscheidung (API+WebUI, LiteLLM nicht fuer M6) #4

Merged
spektr merged 1 commits from m6/openwebui-internal into main 2026-11-11 06:45:19 +01:00
14 changed files with 382 additions and 7 deletions
+13 -4
View File
@@ -88,7 +88,7 @@ Hauptverzeichnisse `llm-server/host/` (Grundkonfiguration des Servers),
| --- | --- | --- |
| KI-Host | Debian 13.6, Deployment-Zugang, Podman 5.4.2, getrennte rootless Konten, Gerätezugriff für beide KI-Konten; IPv4-only (31.10., GNAI-002 6); Lüftersteuerungs-Dienst `gnai-gpu-fan-control` als Dauerlösung (05.11., GNAI-010 10.10); Backup/Restore praktisch belegt (06.11., GNAI-008 14.6) und Backup-Unit per `RequiresMountsFor` boot-sicher (Fix 09.11., GNAI-008 14.6/0.21); Host-Neustart mit automatischem Production-Recovery (06.11. und erneut 09.11.2026, GNAI-008 14.6) | Host-/Treiber-Upgradepfad |
| GPU/ROCm | beide RX 6900 XT als `gfx1030`, `ROCm0` und `ROCm1` im gepinnten Testing-Image erkannt; Modellladen, Layer-Verteilung und reale Inferenz auf beiden GPUs; 64k- und 128k-Stabilität in Testing (GNAI-010 10.5–10.9); GPU-Sensoren im Monitoring mit Alarmgrenzen (GNMON-000 M4/M5, verifiziert 09.11.2026) | Hardware-Ursache des thermischen Defekts an Slot `03:00.0` (GNAI-010 10.10) — Lüfterkurven-Service überbrückt als Dauerlösung |
| LLM-Serving | drei gepinnte Profile (Swift 1.0, Swift 1.5, Unsloth); Testing-Abnahme 64k/128k je Profil (GNAI-010 10.5–10.9); Production-Umgebung live mit Swift 1.5 Q6_K seit 06.11.2026 (Betreiberbeschluss 05.11., GNAI-009 10.4) inklusive Health, Alias und Inferenztest; exklusive Umschaltmechanik mit Recovery (GNAI-008 14.6, T1–T7; nach Neustart 09.11.2026 `gn_ai_active_env{env="production"}=1` verifiziert) | öffentlicher Nutzerpfad (M6), LLM-Metriken jenseits `gn_ai_active_env`, Nutzerakzeptanzprüfung (GNAI-009 5.3) |
| LLM-Serving | drei gepinnte Profile (Swift 1.0, Swift 1.5, Unsloth); Testing-Abnahme 64k/128k je Profil (GNAI-010 10.5–10.9); Production-Umgebung live mit Swift 1.5 Q6_K seit 06.11.2026 (Betreiberbeschluss 05.11., GNAI-009 10.4) inklusive Health, Alias und Inferenztest; exklusive Umschaltmechanik mit Recovery (GNAI-008 14.6, T1–T7; nach Neustart 09.11.2026 `gn_ai_active_env{env="production"}=1` verifiziert); **OpenWebUI v0.13.3 (digest-gepinnt) intern produktiv seit 10.11.2026** — Health 200, Model-Liste, reale Chat-Completion via OpenWebUI → llama.cpp (Rolle `open_webui`, Playbook `webui.yml`, Ingress `10.0.0.90:8081` nur von NPM, Split-DNS `chat.ghostnet.selfhost.eu` live) | öffentlicher Nutzerpfad (M6: NPM-Route + Keycloak-OIDC für chat.*, externe Pos/Neg-Tests), LLM-Metriken jenseits `gn_ai_active_env`, Nutzerakzeptanzprüfung (GNAI-009 5.3) |
| Deployment | lokales Ansible-Control-Projekt und idempotente Hostrollen; Production-Inventar mit Referenz- und Rollback-Profil, `ai_environment_switch` und Backup-Timer deployt und abgenommen (GNAI-008 14.6) | statische CI, Runner, geschützte Freigaben und automatisierter Rollback über die Delivery-Kette |
| VPN | wg-easy v15.4.0; einziger Peer `ghost-laptop` als Server Admin; native nftables-Zielpolicy und NAT, Legacy-Bereinigung, Fremdnetz-, Positiv-/Negativ-, Dienst- und LXC-Wiederanlauftests bestanden; CT 104 am 07.10.2026 als zwölftes Ziel praktisch abgenommen; WAN-Exposure-Erfassung 10.11.2026: nur TCP 80/443 offen, keine AAAA (GNAI-007 7); frischer Stop-Mode-vzdump + hostunabhängige Kopie auf `ghostnet-ki` am 10.11.2026 | Fremdnetz-WAN-Rescan vom Hotspot (UDP 51820, TCP-Portmatrix), FRITZ!Box-Forward-/IPv6-Abgleich, wg-easy-GUI-HTTPS und KI-Server-Test bei eingeschaltetem Host |
| DNS / Pi-hole | CT 100 unter `10.0.0.14` mit digest-gepinntem Pi-hole 2026.09.0 als Podman-Quadlet; SSH/nftables, UDP/TCP-DNS, Split-DNS, Blocking/Allowlist, FRITZ!Box-Cutover, LXC-Reboot und Stop-Mode-Backup am 01.10.2026 abgenommen; Ansible-Projekt am 09.11.2026 adoptiert (ansible-deploy via pct; `site.yml`/`verify-production` `changed=0`) und Split-DNS-Record `mon.ghostnet.selfhost.eu → 10.0.0.10` live (verifiziert per `dig @10.0.0.14`); frischer Stop-Mode-vzdump + hostunabhängige Kopie auf `ghostnet-ki` + praktischer Restore-Test (CT 150) am 10.11.2026 (GNAI-012 10.2) | vollständige App-/OIDC-Regression, praktischer FRITZ!Box-Rollback, mobile DuckDuckGo-/IPv6-Strategie, WireGuard-Profile |
@@ -167,11 +167,20 @@ offenen Erstimplementierungen mehr.
Noch zu entscheiden sind insbesondere:
- API-only oder zusätzlich Weboberfläche als erstes Nutzerprodukt;
- ob LiteLLM in der ersten Veröffentlichung für individuelle Keys, Quoten oder
Routing tatsächlich benötigt wird;
- Freigabe- und Secret-Modell für einen späteren privilegierten Runner.
**Entsieden am 10.11.2026 (Betreiber):** Erstes Nutzerprodukt (M6) ist
**API plus Weboberfläche via OpenWebUI** — OpenWebUI sitzt im
Production-Netz auf `ghostnet-ki` (`10.0.0.90:8081`, nur NPM darf durch),
reicht die Engine über `http://llama-cpp:8080/v1` an und wird öffentlich über
NPM + Keycloak-OIDC freigegeben (`chat.ghostnet.selfhost.eu`). **LiteLLM
fällt für M6 weg** (Entscheidungstor GNAI-001 M6 geschlossen): individuelle
API-Keys und Rollen macht OpenWebUI selbst, Rate-Limits kommen auf
NPM-Ebene; LiteLLM bleibt Zielbild-Kandidat, falls später Quotas/Routing
pro Backend nötig werden. Interner Slice abgenommen 10.11.2026 (Health,
Model-Liste, reale Chat-Completion durch OpenWebUI → llama.cpp mit
Token-Usage).
Bereits entschieden (nicht mehr offen): Das Backupziel ist die bestehende
NFS-Anbindung der Encoding-Pipeline (`pve:/media/datenhdd3/Encoding`), die
Aufbewahrung beträgt 8 Archive und der Restore ist praktisch belegt
+3 -1
View File
@@ -41,7 +41,9 @@ Laufprotokolle noch duplizierte Sollparameter.
- festlegen, ob Sicherheits-Meilenstein 1 den internen LLM-Test blockiert oder
kontrolliert parallel läuft;
- erstes Produkt als API-only oder API plus Weboberfläche festlegen;
- erstes Produkt als API-only oder API plus Weboberfläche festlegen
— **entschieden 10.11.2026: API plus Weboberfläche (OpenWebUI)**, siehe
GNAI-000 Abschnitt 4;
- 64k-Basis und 128k-Ziel als Abnahmemodell bestätigen;
- Verantwortliche für Review, Deployment und Sicherheitsabnahme benennen.
@@ -0,0 +1,14 @@
---
# Deploys the OpenWebUI chat frontend for the Production environment only
# (GNAI-001 milestone 6). The engine stack itself is managed by ai.yml;
# this playbook never changes engine state and requires the Production
# engine network to exist (run ai.yml first).
#
# Usage:
# ansible-playbook -i inventories/production playbooks/webui.yml --check --diff
# ansible-playbook -i inventories/production playbooks/webui.yml
- name: Deploy the OpenWebUI frontend for the Production environment
hosts: llm_servers
become: true
roles:
- role: open_webui
@@ -0,0 +1,19 @@
# open_webui
OpenWebUI chat frontend for the **Production** LLM environment (GNAI-001
milestone 6). Deployed by `playbooks/webui.yml` against
`inventories/production` only.
- Image gepinnt per Digest (`v0.13.3`, `sha256:46b59158…`), `Pull=never`.
- Läuft als `gnai-production` im Production-Netz `gnai-production-llama`
und erreicht die Engine über `http://llama-cpp:8080/v1` — kein Host-Port
der Engine nötig.
- Ingress: `10.0.0.90:8081` (Container-Port 8080); Host-Firewall
(`gnmon_firewall`, Portliste `gnmon_firewall_npm_only_ports`) lässt nur
NPM (`10.0.0.10`) durch.
- Secrets als Podman-Secrets (`type=env`), nie in Git: `WEBUI_SECRET_KEY`
(random, bei Erstanlage erzeugt) und `OPENAI_API_KEY` (Placeholder; die
Engine selbst hat keine Auth).
- Daten: `/var/lib/ghostnet-ai/production/openwebui-data`.
- Interne Phase: lokaler Auth (erster User = Admin). Die öffentliche
OIDC-Phase (NPM-Route + Keycloak-Client) folgt in M6.
@@ -0,0 +1,47 @@
---
# OpenWebUI application role (GNAI-001 milestone 6, first user path).
# Runs as the Production workload account inside the Production engine
# network, so it reaches llama.cpp via the network alias without any host
# port on the engine. The app itself is published on the LAN address only,
# and the host firewall (gnmon_firewall) restricts it to NPM (10.0.0.10).
#
# Image: immutable digest, never a moving tag (GNAI-010 section 3.1).
# v0.13.3 (2026-11-08), index digest resolved 2026-11-11.
open_webui_image: ghcr.io/open-webui/open-webui
open_webui_image_ref: ghcr.io/open-webui/open-webui@sha256:46b59158e6558eeb980bc828e0f1b30238ad24d82ed84d5e5eda1dab2a082f14
open_webui_image_tag: v0.13.3
# Reuses the Production environment identity and network (GNAI-008 7.4).
open_webui_user: gnai-production
open_webui_uid: 980
open_webui_home: /var/lib/ghostnet-ai/production
open_webui_quadlet_dir: /etc/containers/systemd/users/980
open_webui_unit_prefix: gnai-production-openwebui
open_webui_network_name: gnai-production-llama
# Engine endpoint inside the shared Production network. llama.cpp itself
# has no auth; OpenWebUI just needs a non-empty key value, supplied via a
# Podman secret (never in git).
open_webui_engine_base_url: http://llama-cpp:8080/v1
open_webui_engine_api_key_secret: openwebui-engine-api-key
open_webui_webui_secret_key_secret: openwebui-webui-secret-key
# App ingress: LAN address, port 8081, firewall-limited to NPM (CT-103).
open_webui_host_addr: 10.0.0.90
open_webui_host_port: 8081
open_webui_container_port: 8080
# Persistent app data (SQLite DB, uploads, cache).
open_webui_data_dir: /var/lib/ghostnet-ai/production/openwebui-data
# Internal phase: local auth with first-user-is-admin. The public OIDC
# phase (M6 route) switches ENABLE_OAUTH_* on and disables local signup.
open_webui_env:
WEBUI_AUTH: "True"
OPENAI_API_BASE_URL: "{{ open_webui_engine_base_url }}"
DATA_DIR: /app/backend/data
ENABLE_SIGNUP: "True"
DEFAULT_USER_ROLE: user
ENABLE_OAUTH_SIGNUP: "False"
ENABLE_LOGIN_FORM: "True"
ENABLE_RAG_WEB_SEARCH: "False"
@@ -0,0 +1,17 @@
---
- name: Reload workload user manager
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/systemctl
- --user
- daemon-reload
chdir: "{{ open_webui_home }}"
changed_when: false
when: not ansible_check_mode
@@ -0,0 +1,223 @@
---
- name: Verify open_webui role declarations
ansible.builtin.assert:
that:
- open_webui_image_ref is match('^ghcr.io/open-webui/open-webui@sha256:[0-9a-f]{64}$')
- open_webui_network_name == 'gnai-production-llama'
fail_msg: "open_webui must use the pinned digest and the Production network"
- name: Check whether the pinned OpenWebUI image is in the workload store
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/podman
- image
- exists
- "{{ open_webui_image_ref }}"
chdir: "{{ open_webui_home }}"
changed_when: false
failed_when: open_webui_image_present.rc not in [0, 1]
check_mode: false
register: open_webui_image_present
- name: Pull the pinned OpenWebUI image into the workload store
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/podman
- pull
- "{{ open_webui_image_ref }}"
chdir: "{{ open_webui_home }}"
when:
- not ansible_check_mode
- open_webui_image_present.rc != 0
- name: Ensure the OpenWebUI data directory exists
ansible.builtin.file:
path: "{{ open_webui_data_dir }}"
state: directory
owner: "{{ open_webui_user }}"
group: "{{ open_webui_user }}"
mode: "0750"
- name: Check whether the WebUI secret key already exists
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/podman
- secret
- exists
- "{{ open_webui_webui_secret_key_secret }}"
chdir: "{{ open_webui_home }}"
register: open_webui_secret_exists
changed_when: false
failed_when: open_webui_secret_exists.rc not in [0, 1]
check_mode: false
- name: Create the WebUI secret key (random, on the host only)
# Piped via stdin: a root-created temp file would not be readable by the
# workload user (umask), so the secret never touches the filesystem.
ansible.builtin.shell: |
set -euo pipefail
head -c 48 /dev/urandom | base64 \
| /usr/sbin/runuser --user {{ open_webui_user }} -- /usr/bin/env \
HOME={{ open_webui_home }} XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }} \
/usr/bin/podman secret create {{ open_webui_webui_secret_key_secret }} -
# runuser keeps the caller's cwd; the workload user cannot traverse
# /home/ansible-deploy, so run from the workload home.
args:
executable: /bin/bash
chdir: "{{ open_webui_home }}"
when: open_webui_secret_exists.rc == 1
no_log: true
- name: Check whether the engine API key secret already exists
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/podman
- secret
- exists
- "{{ open_webui_engine_api_key_secret }}"
chdir: "{{ open_webui_home }}"
register: open_webui_engine_secret_exists
changed_when: false
failed_when: open_webui_engine_secret_exists.rc not in [0, 1]
check_mode: false
- name: Create the engine API key secret (placeholder, llama.cpp has no auth)
ansible.builtin.shell: |
set -euo pipefail
printf 'local-noauth-%s' "$(head -c 16 /dev/urandom | base64)" \
| /usr/sbin/runuser --user {{ open_webui_user }} -- /usr/bin/env \
HOME={{ open_webui_home }} XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }} \
/usr/bin/podman secret create {{ open_webui_engine_api_key_secret }} -
args:
executable: /bin/bash
chdir: "{{ open_webui_home }}"
when: open_webui_engine_secret_exists.rc == 1
no_log: true
- name: Render the OpenWebUI quadlet
ansible.builtin.template:
src: open-webui.container.j2
dest: "{{ open_webui_quadlet_dir }}/{{ open_webui_unit_prefix }}.container"
owner: root
group: "{{ open_webui_user }}"
mode: "0640"
notify: Reload workload user manager
- name: Ensure the workload user manager knows the rendered unit
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/systemctl
- --user
- daemon-reload
chdir: "{{ open_webui_home }}"
changed_when: false
when: not ansible_check_mode
- name: Read the current state of the OpenWebUI unit
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/systemctl
- --user
- is-active
- "{{ open_webui_unit_prefix }}.service"
chdir: "{{ open_webui_home }}"
changed_when: false
failed_when: open_webui_unit_state.rc not in [0, 3]
check_mode: false
register: open_webui_unit_state
- name: Start the OpenWebUI unit
# Quadlet-generated units are transient and cannot be `enable`d directly;
# the [Install] section of the quadlet handles boot enablement, here we
# only start (same pattern as the inference_engine role).
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/systemctl
- --user
- start
- "{{ open_webui_unit_prefix }}.service"
chdir: "{{ open_webui_home }}"
when:
- not ansible_check_mode
- open_webui_unit_state.rc != 0
- name: Wait for the OpenWebUI health endpoint
ansible.builtin.uri:
url: "http://{{ open_webui_host_addr }}:{{ open_webui_host_port }}/health"
status_code: 200
timeout: 10
register: open_webui_health
retries: 30
delay: 5
until: open_webui_health.status == 200
when: not ansible_check_mode
- name: Verify the engine model list is reachable through OpenWebUI's network
ansible.builtin.command:
argv:
- /usr/sbin/runuser
- --user
- "{{ open_webui_user }}"
- --
- /usr/bin/env
- "HOME={{ open_webui_home }}"
- "XDG_RUNTIME_DIR=/run/user/{{ open_webui_uid }}"
- /usr/bin/podman
- exec
- "{{ open_webui_unit_prefix }}"
- /bin/sh
- -c
- "curl -fsS --max-time 10 {{ open_webui_engine_base_url }}/models | grep -q swift"
chdir: "{{ open_webui_home }}"
register: open_webui_models
changed_when: false
when: not ansible_check_mode
@@ -0,0 +1,26 @@
[Unit]
Description=OpenWebUI chat frontend for the Production LLM engine ({{ open_webui_image_tag }})
Wants={{ open_webui_network_name }}-network.service
After={{ open_webui_network_name }}-network.service
[Container]
Image={{ open_webui_image_ref }}
Pull=never
ContainerName={{ open_webui_unit_prefix }}
Network={{ open_webui_network_name }}
PublishPort={{ open_webui_host_addr }}:{{ open_webui_host_port }}:{{ open_webui_container_port }}
Volume={{ open_webui_data_dir }}:/app/backend/data:rw
# Secrets are injected as environment variables, never written to disk or git.
Secret={{ open_webui_webui_secret_key_secret }},type=env,target=WEBUI_SECRET_KEY
Secret={{ open_webui_engine_api_key_secret }},type=env,target=OPENAI_API_KEY
{% for key, value in open_webui_env.items() %}
Environment={{ key }}={{ value }}
{% endfor %}
[Service]
Restart=on-failure
RestartSec=10
TimeoutStartSec=600
[Install]
WantedBy=default.target
@@ -18,6 +18,13 @@ gnmon_firewall_service_name: nftables.service
gnmon_firewall_dashboard_port: 8080
gnmon_firewall_dashboard_allowed_ip: 10.0.0.10
# LAN-published app ports reachable only from NPM (CT-103): 8080 dashboard
# (GNMON M6), 8081 OpenWebUI (GNAI-001 M6, role open_webui in the llm
# project). NPM terminates TLS and OIDC in front of both.
gnmon_firewall_npm_only_ports:
- 8080
- 8081
# Monitoring APIs bind to 127.0.0.1 (GNMON-000 M5); deny non-loopback
# access to them anyway (defense in depth, GNMON-001 section 7).
gnmon_firewall_loopback_only_ports:
@@ -17,8 +17,11 @@ table inet gnmon_filter {
iifname "lo" accept
# M6: dashboard ingress only from NPM (CT-103, {{ gnmon_firewall_dashboard_allowed_ip }}).
ip saddr != {{ gnmon_firewall_dashboard_allowed_ip }} tcp dport {{ gnmon_firewall_dashboard_port }} drop
# M6: LAN-published app ports (dashboard 8080, OpenWebUI 8081) are
# reachable only from NPM (CT-103, {{ gnmon_firewall_dashboard_allowed_ip }}).
{% for port in gnmon_firewall_npm_only_ports %}
ip saddr != {{ gnmon_firewall_dashboard_allowed_ip }} tcp dport {{ port }} drop
{% endfor %}
# Monitoring APIs are bound to 127.0.0.1; deny non-loopback anyway.
iifname != "lo" tcp dport { {{ gnmon_firewall_loopback_only_ports | join(', ') }} } drop
@@ -32,6 +32,8 @@ dns_pihole_hosts_records:
- "10.0.0.10 auth-admin.ghostnet.selfhost.eu"
# GNMON-001 M6: monitoring dashboard via NPM (CT-103); mirrors the role default.
- "10.0.0.10 mon.ghostnet.selfhost.eu"
# GNAI-001 M6: OpenWebUI via NPM (CT-103).
- "10.0.0.10 chat.ghostnet.selfhost.eu"
dns_pihole_timezone: Europe/Berlin
dns_pihole_database_max_db_days: 30
dns_pihole_secret_name: pihole_web_password
@@ -32,6 +32,8 @@ dns_pihole_hosts_records:
- "10.0.0.10 auth-admin.ghostnet.selfhost.eu"
# GNMON-001 M6: monitoring dashboard via NPM (CT-103); mirrors the role default.
- "10.0.0.10 mon.ghostnet.selfhost.eu"
# GNAI-001 M6: OpenWebUI via NPM (CT-103).
- "10.0.0.10 chat.ghostnet.selfhost.eu"
dns_pihole_timezone: Europe/Berlin
dns_pihole_database_max_db_days: 30
dns_pihole_secret_name: pihole_web_password
@@ -32,6 +32,8 @@ dns_pihole_hosts_records:
- "10.0.0.10 auth-admin.ghostnet.selfhost.eu"
# GNMON-001 M6: monitoring dashboard via NPM (CT-103); mirrors the role default.
- "10.0.0.10 mon.ghostnet.selfhost.eu"
# GNAI-001 M6: OpenWebUI via NPM (CT-103).
- "10.0.0.10 chat.ghostnet.selfhost.eu"
dns_pihole_timezone: Europe/Berlin
dns_pihole_database_max_db_days: 30
dns_pihole_secret_name: pihole_web_password
@@ -13,6 +13,8 @@ dns_pihole_hosts_records:
- "10.0.0.10 auth-admin.ghostnet.selfhost.eu"
# GNMON-001 M6: monitoring dashboard via NPM (CT-103).
- "10.0.0.10 mon.ghostnet.selfhost.eu"
# GNAI-001 M6: OpenWebUI via NPM (CT-103).
- "10.0.0.10 chat.ghostnet.selfhost.eu"
dns_pihole_database_max_db_days: 30
dns_pihole_service_name: pihole.service
dns_pihole_quadlet_path: /etc/containers/systemd/pihole.container