- deploy/compose/infra: Dienst litellm entfernt (D4), Router auf LITELLM_BASE_URL (appsrv08, Port noch offen), alle Images gepinnt, Grafana und pve-exporter ohne Profile, node-exporter für den LXC, Zustand über KNASTER_STATE_DIR - Prometheus: GPU-VMs per file_sd (node, gpu), Proxmox-Knoten per IP, infra-node - Grafana: Datasource und Dashboards provisioniert (grafana.com 10347, 1860, 14574 sowie knaster-Übersicht) - Alertmanager: n8n-Receiver nur vorbereitet (auskommentiert) - Ansible: Rollen infra-stack und gpu-exporters, Playbooks monitoring.yml und monitoring-targets.yml, Skript monitoring-targets.py (tofu output → file_sd) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
co-authored by
Claude Opus 5.5
parent
fd48fd1c7f
commit
bf7c137adf
@@ -0,0 +1,134 @@
|
||||
# Compose-Stack infra: Router, TTS, Postgres, Prometheus, Alertmanager, Grafana, pve-exporter, node-exporter.
|
||||
# Idempotent: Geheimnisse nur anlegen, wenn sie fehlen; compose up ändert nur, was sich geändert hat.
|
||||
- name: Werkzeuge
|
||||
ansible.builtin.apt:
|
||||
name: [rsync, openssl, curl]
|
||||
state: present
|
||||
update_cache: true
|
||||
cache_valid_time: 3600
|
||||
|
||||
- name: Verzeichnisse
|
||||
ansible.builtin.file:
|
||||
path: "{{ item.path }}"
|
||||
state: directory
|
||||
mode: "{{ item.mode }}"
|
||||
loop:
|
||||
- {path: /opt/knaster, mode: "0755"}
|
||||
- {path: "{{ infra_repo_dir }}", mode: "0755"}
|
||||
- {path: "{{ infra_state_dir }}", mode: "0700"}
|
||||
- {path: "{{ infra_state_dir }}/secrets", mode: "0700"}
|
||||
|
||||
# Nur die Teile, die Build und Konfiguration brauchen. Nie .git, .venv, tofu-State oder .env-Dateien.
|
||||
- name: Repo-Auszug von ops abgleichen
|
||||
ansible.posix.synchronize:
|
||||
src: "{{ infra_repo_src }}/"
|
||||
dest: "{{ infra_repo_dir }}/"
|
||||
delete: true
|
||||
rsync_opts:
|
||||
- "--exclude=__pycache__"
|
||||
- "--exclude=.pytest_cache"
|
||||
- "--exclude=*.egg-info"
|
||||
- "--exclude=.env"
|
||||
- "--exclude=secrets/"
|
||||
- "--include=/core/***"
|
||||
- "--include=/config/***"
|
||||
- "--include=/catalog/***"
|
||||
- "--include=/skills/***"
|
||||
- "--include=/sites/***"
|
||||
- "--include=/deploy/***"
|
||||
- "--exclude=*"
|
||||
|
||||
- name: Einstellungen in .env
|
||||
ansible.builtin.lineinfile:
|
||||
path: "{{ infra_state_dir }}/.env"
|
||||
regexp: "^{{ item.key }}="
|
||||
line: "{{ item.key }}={{ item.value }}"
|
||||
create: true
|
||||
mode: "0600"
|
||||
loop: "{{ infra_settings | dict2items }}"
|
||||
|
||||
# Gibt nur Namen aus, nie Werte
|
||||
- name: Fehlende Geheimnisse im Container erzeugen
|
||||
ansible.builtin.shell: |
|
||||
set -eu
|
||||
umask 077
|
||||
f={{ infra_state_dir }}/.env
|
||||
for k in {{ infra_secrets | join(' ') }}; do
|
||||
grep -q "^$k=" "$f" || { echo "$k=$(openssl rand -hex 24)" >> "$f"; echo "neu: $k"; }
|
||||
done
|
||||
chmod 0600 "$f"
|
||||
args:
|
||||
executable: /bin/bash
|
||||
register: secrets_gen
|
||||
changed_when: "'neu:' in secrets_gen.stdout"
|
||||
|
||||
- name: Router-Token für Alertmanager (= KNASTER_API_TOKEN, lesbar für nobody im Container)
|
||||
ansible.builtin.shell: |
|
||||
set -eu
|
||||
umask 077
|
||||
cd {{ infra_state_dir }}
|
||||
v=$(sed -n 's/^KNASTER_API_TOKEN=//p' .env)
|
||||
t=secrets/router_token
|
||||
if [ ! -f "$t" ] || [ "$(cat "$t")" != "$v" ]; then printf '%s' "$v" > "$t"; echo geändert; fi
|
||||
chown 65534:65534 "$t"
|
||||
chmod 0400 "$t"
|
||||
args:
|
||||
executable: /bin/bash
|
||||
register: router_token
|
||||
changed_when: "'geändert' in router_token.stdout"
|
||||
no_log: true
|
||||
|
||||
- name: Proxmox-Lese-Token für den pve-exporter
|
||||
ansible.builtin.copy:
|
||||
src: "{{ infra_pve_exporter_env }}"
|
||||
dest: "{{ infra_state_dir }}/secrets/pve-exporter.env"
|
||||
mode: "0600"
|
||||
no_log: true
|
||||
|
||||
- name: file_sd-Ziele
|
||||
ansible.builtin.include_tasks: targets.yml
|
||||
|
||||
- name: Stack bauen und starten
|
||||
ansible.builtin.command: >
|
||||
docker compose -p {{ infra_compose_project }} --env-file {{ infra_state_dir }}/.env
|
||||
-f {{ infra_compose_file }} up -d --build --remove-orphans
|
||||
register: compose_up
|
||||
changed_when: compose_up.stderr is search('Container \S+\s+(Created|Recreated|Started)')
|
||||
|
||||
- name: Prometheus und Alertmanager laden geänderte Konfiguration (bind-mount, kein Neustart nötig)
|
||||
ansible.builtin.command: "curl -fsS -m 10 -X POST http://127.0.0.1:{{ item }}/-/reload"
|
||||
loop: [9090, 9093]
|
||||
register: reload
|
||||
retries: 15
|
||||
delay: 4
|
||||
until: reload.rc == 0
|
||||
changed_when: false
|
||||
|
||||
- name: Grafana-Konto kiosk (Viewer) anlegen, Passwort aus .env
|
||||
ansible.builtin.shell: |
|
||||
set -eu
|
||||
cd {{ infra_state_dir }}
|
||||
A=$(sed -n 's/^GRAFANA_ADMIN_PASSWORD=//p' .env)
|
||||
K=$(sed -n 's/^GRAFANA_KIOSK_PASSWORD=//p' .env)
|
||||
umask 077
|
||||
cfg=$(mktemp)
|
||||
trap 'rm -f "$cfg"' EXIT
|
||||
printf 'user = "admin:%s"\n' "$A" > "$cfg" # Passwort nicht in der Prozessliste
|
||||
g() { curl -sS -K "$cfg" -H 'Content-Type: application/json' "$@"; }
|
||||
for i in $(seq 60); do curl -fsS -m 3 http://127.0.0.1:3000/api/health >/dev/null 2>&1 && break; sleep 2; done
|
||||
code=$(g -o /dev/null -w '%{http_code}' 'http://127.0.0.1:3000/api/users/lookup?loginOrEmail=kiosk')
|
||||
if [ "$code" = 404 ]; then
|
||||
printf '{"name":"Kiosk","login":"kiosk","password":"%s"}' "$K" \
|
||||
| g -f -X POST --data @- http://127.0.0.1:3000/api/admin/users >/dev/null
|
||||
echo angelegt
|
||||
elif [ "$code" != 200 ]; then
|
||||
echo "Grafana-API antwortet $code" >&2
|
||||
exit 1
|
||||
fi
|
||||
id=$(g -f 'http://127.0.0.1:3000/api/users/lookup?loginOrEmail=kiosk' | python3 -c 'import json,sys; print(json.load(sys.stdin)["id"])')
|
||||
g -f -X PATCH --data '{"role":"Viewer"}' "http://127.0.0.1:3000/api/org/users/$id" >/dev/null
|
||||
args:
|
||||
executable: /bin/bash
|
||||
register: kiosk
|
||||
changed_when: "'angelegt' in kiosk.stdout"
|
||||
no_log: true
|
||||
@@ -0,0 +1,26 @@
|
||||
# file_sd-Ziele der GPU-VMs. Auch einzeln aufrufbar: playbooks/monitoring-targets.yml
|
||||
# (nach refresh-vm-access.sh, wenn sich DHCP-Adressen geändert haben). Prometheus liest die Dateien selbst neu.
|
||||
- name: file_sd-Ziele aus tofu output erzeugen (auf ops)
|
||||
ansible.builtin.shell: |
|
||||
set -euo pipefail
|
||||
tofu output -json gpu_vms | python3 {{ infra_repo_src }}/platform/proxmox/scripts/monitoring-targets.py \
|
||||
--registry {{ infra_repo_src }}/sites/{{ knaster_site }}/registry.yaml --out {{ infra_targets_dir }}
|
||||
args:
|
||||
chdir: "{{ infra_tofu_dir }}"
|
||||
executable: /bin/bash
|
||||
delegate_to: localhost
|
||||
run_once: true
|
||||
changed_when: false
|
||||
|
||||
- name: file_sd-Verzeichnis
|
||||
ansible.builtin.file:
|
||||
path: "{{ infra_state_dir }}/file_sd"
|
||||
state: directory
|
||||
mode: "0755"
|
||||
|
||||
- name: file_sd-Ziele in den Container kopieren
|
||||
ansible.builtin.copy:
|
||||
src: "{{ infra_targets_dir }}/{{ item }}.json"
|
||||
dest: "{{ infra_state_dir }}/file_sd/{{ item }}.json"
|
||||
mode: "0644"
|
||||
loop: [node, gpu]
|
||||
Reference in New Issue
Block a user