Add shared nginx-proxy role; deploy Hermes in Docker; rename Paperclip

nginx-proxy (new role):
- Shared nginx + certbot under /opt/nginx, replacing the copies that
  lived inside the Paperclip compose project
- Apps publish via tasks_from: add_vhost (cert marker -> issue cert ->
  write vhost -> nginx -t -> reload); a vhost that fails nginx -t is
  rolled back
- Certs issued one per domain via Cloudflare DNS challenge; domains
  registered as marker files in cert-domains/; optional LE staging
- Stock nginx.conf plus conf.d (directory mounts, not single-file
  mounts); catch-all server drops unknown Host/SNI
- Containers renamed to nginx and certbot

hermes_docker (new role):
- Official Hermes image, host networking, gateway and dashboard bound
  to loopback and published through nginx-proxy
- Dashboard login via basic-auth provider, secrets from the vault
- Main model set from hermes_model (fantasmic endpoint, key from vault);
  config.yaml overrides are merged from a 0600 staged file that is
  removed afterwards, so the key never appears in a command line
- Data dir owner-only (HERMES_HOME_MODE=0700)

paperclip-docker:
- Remove nginx, certbot and docker-compose download tasks; deploy with
  community.docker.docker_compose_v2 (idempotent, remove_orphans)
- Stop passing the Cloudflare token and certbot email to the Paperclip
  container via env_file
- Bind Postgres to 127.0.0.1 (host networking exposed it on all
  interfaces); stop chowning pgdata, which the postgres image owns
- Handler recreates the paperclip service so .env changes apply
- Domain m-v.wdw-labs.cloud -> cranium-command.wdw-labs.cloud

Playbooks:
- Add day1_deploy_nginx_proxy.yml and day1_deploy_hermes_docker.yml
- deploy_paperclip_stack.yml: run nginx-proxy before paperclip-docker,
  CLI-based container checks (no Docker SDK dependency)
- Remove deploy-hermes.yml

Host vars: add nginx_proxy_certbot_email and nginx_proxy_certbot_staging
(staging stays on until the switch to real certs).
This commit is contained in:
rblundon committed 2026-09-30 23:12:40 -05:00
1 parent 5b48536c32
commit 881e2e24a8
37 files changed
+1683 -755

No files matched your search

+97 -110
View File
@@ -1,27 +1,6 @@
---
- name: Install Docker SDK for Python
apt:
name: python3-docker
state: present
become: true
- name: Download latest docker-compose from GitHub
shell: |
curl -L "https://github.com/docker/compose/releases/latest/download/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose
chmod +x /usr/local/bin/docker-compose
become: true
- name: Verify docker-compose installation
shell: /usr/local/bin/docker-compose --version
register: compose_version
become: true
- name: Show docker-compose version
debug:
msg: "{{ compose_version.stdout }}"
- name: Create paperclip directories
file:
ansible.builtin.file:
path: "{{ item }}"
state: directory
mode: '0755'
@@ -29,97 +8,77 @@
group: root
loop:
- "{{ paperclip_base_path }}"
- "{{ paperclip_data_dir }}"
- "{{ paperclip_certs_dir }}"
# The postgres image chowns its own data dir on start; do not chown it here.
- "{{ paperclip_base_path }}/pgdata"
become: true
tags: [paperclip, files]
- name: Create OpenCode config directory
file:
path: "{{ paperclip_base_path }}/.config/opencode"
# Container runs as UID 1000. Deliberately NOT recursive: recurse + mode would
# chmod every file the app writes under these paths on each run. A fresh
# deploy has no root-owned content to repair; if you ever need to, run a
# one-off `chown -R 1000:1000 <dir>` by hand.
- name: Create container-owned directories (UID 1000)
ansible.builtin.file:
path: "{{ item }}"
state: directory
mode: '0755'
owner: 1000
group: 1000
loop:
- "{{ paperclip_data_dir }}"
- "{{ paperclip_base_path }}/.config/opencode"
- "{{ paperclip_base_path }}/.local/share/opencode"
become: true
tags: [paperclip, files]
- name: Create OpenCode .local/share directory
file:
path: "{{ paperclip_base_path }}/.local/share/opencode"
state: directory
mode: '0755'
owner: 1000
group: 1000
become: yes
- name: Deploy OpenCode auth.json with vLLM credentials
template:
src: opencode-auth.json.j2
dest: "{{ paperclip_base_path }}/.local/share/opencode/auth.json"
mode: '0600'
owner: 1000
group: 1000
become: true
no_log: true
- name: Fix directory ownership for container user (UID 1000)
shell: |
chown -R 1000:1000 {{ paperclip_data_dir }}
chown -R 1000:1000 {{ paperclip_base_path }}/pgdata
chown -R 1000:1000 {{ paperclip_base_path }}/.config
become: true
# - name: Deploy OpenCode auth.json with vLLM credentials
# template:
# src: opencode-auth.json.j2
# dest: "{{ paperclip_base_path }}/.local/share/opencode/auth.json"
# mode: '0600'
# owner: 1000
# group: 1000
# become: true
# no_log: true
# This file is both the compose interpolation source and the container's
# env_file, so everything here is visible inside the Paperclip container.
# Infrastructure secrets (Cloudflare token, certbot email) live in
# /opt/nginx/.env instead.
- name: Write secrets and configuration to .env
copy:
ansible.builtin.copy:
content: |
# Authentication secrets (from vault)
BETTER_AUTH_SECRET={{ vault_better_auth_secret }}
PAPERCLIP_AGENT_JWT_SECRET={{ vault_paperclip_agent_jwt_secret }}
# Infrastructure configuration
CLOUDFLARE_API_TOKEN={{ vault_cloudflare_api_token }}
CERTBOT_EMAIL={{ certbot_email }}
PAPERCLIP_DOMAIN={{ paperclip_domain }}
VLLM_ENDPOINT={{ vllm_endpoint }}
VLLM_API_KEY={{ vault_vllm_api_key }}
# Inference endpoints
ILLUMINATIONS_ENDPOINT={{ illuminations_endpoint }}
ILLUMINATIONS_API_KEY={{ vault_illuminations_api_key }}
FANTASMIC_ENDPOINT={{ fantasmic_endpoint }}
FANTASMIC_API_KEY={{ vault_fantasmic_api_key }}
dest: "{{ paperclip_base_path }}/.env"
mode: '0600'
owner: root
group: root
become: true
no_log: true
notify: Restart paperclip stack
tags: [paperclip, files]
# No notify: the deploy task below recreates whatever the compose file changed.
- name: Template docker-compose.yml
template:
ansible.builtin.template:
src: docker-compose.yml.j2
dest: "{{ paperclip_base_path }}/docker-compose.yml"
mode: '0644'
owner: root
group: root
become: true
notify: Restart paperclip stack
- name: Template nginx config
template:
src: nginx.conf.j2
dest: "{{ paperclip_base_path }}/nginx.conf"
mode: '0644'
owner: root
group: root
become: true
notify: Reload nginx
- name: Template certbot init script
template:
src: certbot-init.sh.j2
dest: "{{ paperclip_base_path }}/certbot-init.sh"
mode: '0755'
owner: root
group: root
become: true
tags: [paperclip, files]
- name: Template OpenCode config
template:
ansible.builtin.template:
src: opencode.jsonc.j2
dest: "{{ paperclip_base_path }}/.config/opencode/opencode.jsonc"
mode: '0644'
@@ -127,43 +86,71 @@
group: 1000
become: true
notify: Restart paperclip stack
tags: [paperclip, files]
- name: Stop and remove old containers
shell: |
cd {{ paperclip_base_path }}
/usr/local/bin/docker-compose down --remove-orphans 2>/dev/null || true
# Idempotent: only changed services are recreated. remove_orphans also cleans
# up the old paperclip-nginx / paperclip-certbot containers if they still exist.
- name: Deploy paperclip stack
community.docker.docker_compose_v2:
project_src: "{{ paperclip_base_path }}"
pull: missing
remove_orphans: true
state: present
register: paperclip_deploy
become: true
ignore_errors: true
- name: Deploy docker compose stack
shell: |
cd {{ paperclip_base_path }}
/usr/local/bin/docker-compose up -d
become: true
register: compose_result
tags: [paperclip, deploy]
- name: Wait for Paperclip to be ready
uri:
ansible.builtin.uri:
url: "http://localhost:{{ paperclip_port }}/health"
status_code: 200
register: result
until: result.status == 200
register: paperclip_health
until: paperclip_health.status == 200
retries: 30
delay: 5
ignore_errors: true
tags: [paperclip, deploy, verify]
- name: Wait for nginx to be ready
uri:
url: "https://{{ paperclip_domain }}/"
validate_certs: no
status_code: [200, 301, 302]
register: result
until: result.status in [200, 301, 302]
retries: 30
- name: Warn if Paperclip never became healthy
ansible.builtin.debug:
msg: "WARNING: Paperclip did not report healthy on port {{ paperclip_port }}. Check: docker logs paperclip"
when: paperclip_health.status | default(0) != 200
tags: [paperclip, verify]
# Requires day1_deploy_nginx_proxy.yml to have run first.
- name: Publish Paperclip through nginx-proxy
ansible.builtin.include_role:
name: nginx-proxy
tasks_from: add_vhost
apply:
become: true
tags: [paperclip, nginx]
vars:
vhost_name: paperclip
vhost_domain: "{{ paperclip_domain }}"
vhost_upstream_port: "{{ paperclip_port }}"
vhost_health_path: /health
tags: [paperclip, nginx]
# --resolve keeps this independent of DNS: it tests nginx, the cert and the
# upstream. DNS is a separate (manual) prerequisite.
- name: Verify HTTPS through nginx
ansible.builtin.command: >-
curl -sk -o /dev/null -w '%{http_code}'
--resolve {{ paperclip_domain }}:443:127.0.0.1
https://{{ paperclip_domain }}/
register: paperclip_https
until: paperclip_https.stdout in ['200', '301', '302']
retries: 6
delay: 5
ignore_errors: true
changed_when: false
failed_when: false
tags: [paperclip, verify]
- name: Verify Paperclip is accessible via https
debug:
msg: "Paperclip is accessible at https://{{ paperclip_domain }}"
when: result.status is defined and result.status in [200, 301, 302]
- name: Report HTTPS result
ansible.builtin.debug:
msg: >-
{{ 'Paperclip is serving at https://' ~ paperclip_domain
if paperclip_https.stdout in ['200', '301', '302']
else 'WARNING: https://' ~ paperclip_domain ~ ' returned ' ~ (paperclip_https.stdout | default('no response')) ~ ' (502 usually means Paperclip is not up yet)' }}
tags: [paperclip, verify]