Compare commits

...

3 Commits

Author SHA1 Message Date
843347ad00 feat(tls): Let's Encrypt via OVH DNS-01 + host nginx vhost
tooling/issue-letsencrypt.sh: manual script, issues one cert per name
(apex + LE_SUBDOMAINS) with acme.sh + OVH DNS-01 (no inbound ports),
installs to certs/<name>/ and reloads nginx. nginx/ulicraft.conf.tmpl:
TLS vhosts for apex/pack (static) + auth (proxy to drasl). OVH creds and
LE_EMAIL in .env.example; certs/ gitignored. plan/15-letsencrypt.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 23:35:07 +02:00
4c874e20ed refactor(ingress): extract caddy to its own compose file; add nginx path
Caddy is no longer in the base stack. Two ingress paths now:
- docker-compose.caddy.yml — LAN/air-gap; build-stack.sh adds it for all
  up modes + the prep pre-bake. Holds the mcnet auth./pack. aliases.
- docker-compose.nginx.yml — production behind the host's nginx: publishes
  drasl on 127.0.0.1:25585, points minecraft at the host (extra_hosts) over
  HTTPS for pack + authlib.

Base loses the caddy service and minecraft's caddy depends_on (re-added by
caddy.yml). Blessing Skin override now requires caddy.yml (it only swaps
caddy's core conf).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 23:35:07 +02:00
67718d8e05 docs(deploy): cochi redeploy guide + deploy skill
plan/14-deploy.md: git pull main + hard restart (compose down ->
build-stack.sh --up online), drasl/online, verify + rollback.
.claude/skills/deploy: skill to run it (pre-check, pull, hard restart,
verify) with prod guardrails.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-07 23:22:37 +02:00
12 changed files with 528 additions and 41 deletions

View File

@@ -0,0 +1,66 @@
---
name: deploy
description: Redeploy the Ulicraft Minecraft stack to the production host `cochi` — git pull on main then a hard restart (compose down + build-stack.sh --up online). Use when the user says "deploy", "redeploy", "deploy to cochi", "push to prod", "ship it", or "restart the server" for this project. Causes brief Minecraft downtime.
---
# Deploy Ulicraft to `cochi`
Redeploy the running stack on the production host. Full reference:
`plan/14-deploy.md`.
## Fixed parameters
- Host: `cochi` (SSH alias)
- Path: `/home/ubuntu/mc/ulicraft-server-v1`
- Branch: `main`
- Run mode: `online` → compose files `docker-compose.yml` + `docker-compose.static.yml`
- Auth: Drasl (base stack)
- Restart: **hard** (down → up) — players are disconnected for a few minutes
## Procedure
This is an outward-facing, downtime-causing action on production. The user
invoking this skill is the authorization to proceed, but:
1. **Pre-check (read-only).** See what would change before tearing anything down:
```bash
ssh cochi 'cd /home/ubuntu/mc/ulicraft-server-v1 && git fetch && git log --oneline HEAD..@{u}'
```
Report the incoming commits. If the list is empty, tell the user the host is
already up to date and ask whether to restart anyway (they may want a plain
restart). If `git fetch` fails (host unreachable, auth), STOP and report.
2. **Pull + hard restart** (single SSH session, halts on any error):
```bash
ssh cochi 'set -e
cd /home/ubuntu/mc/ulicraft-server-v1
git pull --ff-only
docker compose -f docker-compose.yml -f docker-compose.static.yml down --remove-orphans
tooling/build-stack.sh --up online'
```
- `--ff-only` refuses to merge — if the pull is not a fast-forward, STOP and
report (local changes on the host need manual resolution).
- `build-stack.sh --up online` renders configs (needs `.env` complete:
`BASE_DOMAIN`, `HOST_LAN_IP`, `RCON_PASSWORD`) and runs `compose up -d`.
3. **Verify.**
```bash
ssh cochi 'cd /home/ubuntu/mc/ulicraft-server-v1 && docker compose -f docker-compose.yml -f docker-compose.static.yml ps'
```
Confirm `caddy`, `drasl`, `minecraft`, `mc-backup` are Up. Tail the minecraft
log briefly and confirm it reaches `Done` (server ready):
```bash
ssh cochi 'cd /home/ubuntu/mc/ulicraft-server-v1 && timeout 120 docker compose -f docker-compose.yml -f docker-compose.static.yml logs -f minecraft' 2>/dev/null | grep -m1 "Done"
```
## Guardrails
- Never `git reset --hard` or discard changes on `cochi` without telling the user
first — there may be host-local edits.
- Volumes (`mc_data`, `drasl_state`) persist across `down`; the world is safe. Do
NOT pass `-v`/`--volumes` to `down`.
- If a step fails, STOP and surface the exact error + the failing command. Do not
improvise recovery on production.
- Blessing Skin variant or `airgap` mode changes the compose file set — if the
user asks for those, follow `plan/14-deploy.md` instead of the fixed commands
above.

View File

@@ -18,6 +18,19 @@ RCON_PASSWORD=change-me-to-something-random
# Only needed if using CurseForge-exclusive mods:
# CF_API_KEY=your-curseforge-api-key
# ── Let's Encrypt via OVH DNS-01 (tooling/issue-letsencrypt.sh) ───────
# Only needed to issue real TLS certs. DNS-01 needs no inbound ports.
LE_EMAIL=you@example.com
# space-separated subdomains; apex ${BASE_DOMAIN} is always included
LE_SUBDOMAINS=auth pack
# OVH API creds (DNS zone write). Create at https://eu.api.ovh.com/createToken/
# OVH_CK can be blank on first run — acme.sh prints an auth URL, then paste it.
OVH_END_POINT=ovh-eu
OVH_AK=
OVH_AS=
OVH_CK=
# LE_STAGING=1 # use the untrusted staging CA for dry runs
# ── Blessing Skin auth variant (docker-compose.blessingskin.yml) ──────
# Only needed when running the Blessing Skin override instead of Drasl.
BS_DB_NAME=blessingskin

3
.gitignore vendored
View File

@@ -24,3 +24,6 @@ landing/dist/
# exported Caddy CA (per-deploy artifact)
caddy/caddy-root-ca.crt
# Let's Encrypt certs + private keys (issued by tooling/issue-letsencrypt.sh)
certs/

View File

@@ -1,9 +1,11 @@
# ──────────────────────────────────────────────────────────────────
# Blessing Skin auth variant — use INSTEAD of Drasl.
# ──────────────────────────────────────────────────────────────────
# Layer this over the base file:
# Layer this over the base file AND the caddy ingress (caddy lives in its own
# file now; this override only swaps caddy's core conf, so caddy.yml is required):
#
# docker compose -f docker-compose.yml -f docker-compose.blessingskin.yml up -d
# docker compose -f docker-compose.yml -f docker-compose.caddy.yml \
# -f docker-compose.blessingskin.yml up -d
#
# What it does vs the base stack:
# - drasl → left running but UNUSED + unreachable (Caddy no longer

46
docker-compose.caddy.yml Normal file
View File

@@ -0,0 +1,46 @@
# ──────────────────────────────────────────────────────────────────
# Caddy ingress — LAN / air-gap path. NOT started with the base stack.
# ──────────────────────────────────────────────────────────────────
# Caddy was moved out of docker-compose.yml so the base stack can run behind
# the host's nginx (production) without it. build-stack.sh adds this file for
# its online/airgap/core modes; static/mirror overrides mount extra vhosts in.
#
# docker compose -f docker-compose.yml -f docker-compose.caddy.yml up -d
#
# Holds the mcnet aliases (auth./pack.${BASE_DOMAIN}) so containers resolve the
# stack's own names internally — this is why minecraft can reach the pack here
# without extra_hosts (the nginx path uses extra_hosts instead).
services:
caddy:
image: caddy:alpine
container_name: caddy
restart: unless-stopped
ports:
- "80:80"
environment:
BASE_DOMAIN: ${BASE_DOMAIN}
volumes:
- ./caddy/Caddyfile:/etc/caddy/Caddyfile:ro
- ./caddy/conf.d/00-core.caddy:/etc/caddy/conf.d/00-core.caddy:ro
- ./caddy/conf.d/10-static.caddy:/etc/caddy/conf.d/10-static.caddy:ro
- ./www:/srv/www:ro
- ./mirror/launcher:/srv/launcher:ro
- ./caddy/caddy-root-ca.crt:/srv/ca-pub/ca.crt:ro
- ./pack:/srv/pack:ro
# Custom (locally-hosted) mod jars live OUTSIDE ./pack so packwiz refresh
# doesn't index them as direct files. Served at pack.${BASE_DOMAIN}/custom/.
- ./custom:/srv/pack/custom:ro
networks:
mcnet:
# The upstream spoof-host aliases live in docker-compose.mirror.yml —
# adding them here would hijack maven.neoforged.net etc. for ALL mcnet
# containers, breaking the ONLINE NeoForge install during pre-bake.
aliases:
- "auth.${BASE_DOMAIN}"
- "pack.${BASE_DOMAIN}"
# Restore the ordering the base file used to have (base drops it so it can run
# caddy-less behind nginx).
minecraft:
depends_on:
- caddy

30
docker-compose.nginx.yml Normal file
View File

@@ -0,0 +1,30 @@
# ──────────────────────────────────────────────────────────────────
# Production ingress = the HOST's nginx + Let's Encrypt. No caddy.
# ──────────────────────────────────────────────────────────────────
# docker compose -f docker-compose.yml -f docker-compose.nginx.yml up -d
#
# nginx (on the host, already installed) terminates TLS with the LE certs from
# tooling/issue-letsencrypt.sh and:
# - serves apex + pack.${BASE_DOMAIN} as static files (./www, ./pack, ./custom)
# - reverse-proxies auth.${BASE_DOMAIN} -> 127.0.0.1:25585 (drasl)
# See nginx/ulicraft.conf.tmpl.
#
# This override:
# - publishes drasl on localhost so the host nginx can reach it
# - points the minecraft container at the host (extra_hosts) over HTTPS, so it
# fetches the pack and validates sessions through the same public names the
# LE certs cover (JVM trusts Let's Encrypt out of the box)
services:
drasl:
ports:
- "127.0.0.1:25585:25585"
minecraft:
# Make the public names resolve to the host running nginx.
extra_hosts:
- "auth.${BASE_DOMAIN}:host-gateway"
- "pack.${BASE_DOMAIN}:host-gateway"
environment:
# Use HTTPS now that nginx terminates TLS with a publicly-trusted cert.
JVM_OPTS: "-javaagent:/extras/authlib-injector.jar=https://auth.${BASE_DOMAIN}/authlib-injector"
PACKWIZ_URL: "https://pack.${BASE_DOMAIN}/pack.toml"

View File

@@ -5,45 +5,21 @@
# DNS is provided by YOUR party DNS server — see tooling/dns-records.sh for the
# records to add (online vs airgap). A turnkey dnsmasq is available as an
# optional layer: docker-compose.dnsmasq.yml.
# - caddy is the only HTTP ingress; holds mcnet aliases so the stack
# resolves auth.${BASE_DOMAIN} / packwiz.${BASE_DOMAIN} internally too
# - ingress is layered on, NOT in this base file: caddy (docker-compose.caddy.yml,
# LAN/air-gap, holds mcnet aliases) OR the host's nginx (docker-compose.nginx.yml
# + Let's Encrypt, production)
# - drasl handles auth (password login; NO Keycloak/OIDC in this stack)
# - packwiz pack is served by caddy at packwiz.${BASE_DOMAIN}
# - packwiz pack is served by the ingress at pack.${BASE_DOMAIN}
# - minecraft installs mods via PACKWIZ_URL, authlib-injector -> auth.
# ──────────────────────────────────────────────────────────────────
services:
# Core ingress. Static site + air-gap mirror are layered in via override
# files (docker-compose.static.yml / .mirror.yml) — see build-stack.sh --up.
caddy:
image: caddy:alpine
container_name: caddy
restart: unless-stopped
ports:
- "80:80"
environment:
BASE_DOMAIN: ${BASE_DOMAIN}
volumes:
- ./caddy/Caddyfile:/etc/caddy/Caddyfile:ro
- ./caddy/conf.d/00-core.caddy:/etc/caddy/conf.d/00-core.caddy:ro
- ./caddy/conf.d/10-static.caddy:/etc/caddy/conf.d/10-static.caddy:ro
- ./www:/srv/www:ro
- ./mirror/launcher:/srv/launcher:ro
- ./caddy/caddy-root-ca.crt:/srv/ca-pub/ca.crt:ro
- ./pack:/srv/pack:ro
# Custom (locally-hosted) mod jars live OUTSIDE ./pack so packwiz refresh
# doesn't index them as direct files — they're external download targets
# referenced by pack/mods/*.pw.toml. Served at pack.${BASE_DOMAIN}/custom/.
- ./custom:/srv/pack/custom:ro
networks:
mcnet:
# Only the stack's own subdomains in the base. The upstream spoof-host
# aliases live in docker-compose.mirror.yml — adding them here would
# hijack maven.neoforged.net / libraries.minecraft.net for ALL mcnet
# containers, breaking the ONLINE NeoForge install during pre-bake.
aliases:
- "auth.${BASE_DOMAIN}"
- "pack.${BASE_DOMAIN}"
# NOTE: the HTTP(S) ingress is NOT in this base file.
# - LAN / air-gap: caddy (docker-compose.caddy.yml) — added by build-stack.sh
# - production: the HOST's nginx + Let's Encrypt (docker-compose.nginx.yml
# publishes drasl to localhost; nginx/ulicraft.conf.tmpl)
# Base alone (drasl + minecraft + mc-backup) is not a complete deployment —
# always layer one ingress file on top.
drasl:
image: unmojang/drasl:latest
container_name: drasl
@@ -95,7 +71,9 @@ services:
ENFORCE_SECURE_PROFILE: "FALSE" # required for authlib-injector on 1.21+
ALLOW_FLIGHT: "TRUE" # many tech mods need this
# ── Mod source: packwiz pack served by caddy ────────────────
# ── Mod source: packwiz pack served by the ingress ──────────
# http + caddy alias by default; the nginx override switches this to
# https + host-gateway (extra_hosts) for the production path.
PACKWIZ_URL: "http://pack.${BASE_DOMAIN}/pack.toml"
# ── RCON for remote admin ───────────────────────────────────
@@ -108,8 +86,7 @@ services:
- mc_data:/data
- ./runtime:/extras:ro # authlib-injector.jar lives here
depends_on:
- drasl
- caddy # packwiz pack is served by caddy
- drasl # ingress (caddy/nginx) is layered separately
networks:
- mcnet

83
nginx/ulicraft.conf.tmpl Normal file
View File

@@ -0,0 +1,83 @@
# ──────────────────────────────────────────────────────────────────
# Ulicraft host nginx — TLS ingress with Let's Encrypt certs.
# ──────────────────────────────────────────────────────────────────
# Template. Render with BOTH vars then install:
#
# APP_DIR=/home/ubuntu/mc/ulicraft-server-v1 BASE_DOMAIN=ulicraft.net \
# envsubst '$APP_DIR $BASE_DOMAIN' \
# < nginx/ulicraft.conf.tmpl | sudo tee /etc/nginx/sites-available/ulicraft.conf
# sudo ln -sf /etc/nginx/sites-available/ulicraft.conf /etc/nginx/sites-enabled/
# sudo nginx -t && sudo systemctl reload nginx
#
# Prereqs:
# - certs issued: tooling/issue-letsencrypt.sh -> $APP_DIR/certs/<name>/{cert,key}.pem
# - the docker stack up with the nginx override (publishes drasl on 127.0.0.1:25585):
# docker compose -f docker-compose.yml -f docker-compose.nginx.yml up -d
# - nginx's user (www-data) can READ $APP_DIR/{www,pack,custom,certs}
# ──────────────────────────────────────────────────────────────────
# HTTP -> HTTPS for every name.
server {
listen 80;
listen [::]:80;
server_name ${BASE_DOMAIN} auth.${BASE_DOMAIN} pack.${BASE_DOMAIN};
return 301 https://$host$request_uri;
}
# Apex — static landing page.
server {
listen 443 ssl;
listen [::]:443 ssl;
http2 on;
server_name ${BASE_DOMAIN};
ssl_certificate ${APP_DIR}/certs/${BASE_DOMAIN}/cert.pem;
ssl_certificate_key ${APP_DIR}/certs/${BASE_DOMAIN}/key.pem;
root ${APP_DIR}/www;
index index.html;
location / { try_files $uri $uri/ =404; }
}
# Pack — packwiz metadata (./pack) + custom jars (./custom at /custom/).
server {
listen 443 ssl;
listen [::]:443 ssl;
http2 on;
server_name pack.${BASE_DOMAIN};
ssl_certificate ${APP_DIR}/certs/pack.${BASE_DOMAIN}/cert.pem;
ssl_certificate_key ${APP_DIR}/certs/pack.${BASE_DOMAIN}/key.pem;
# Locally-hosted custom mod jars live outside ./pack; expose them at /custom/.
location /custom/ {
alias ${APP_DIR}/custom/;
autoindex on;
}
location / {
root ${APP_DIR}/pack;
autoindex on;
}
}
# Auth — reverse proxy to drasl (published on localhost by the nginx override).
server {
listen 443 ssl;
listen [::]:443 ssl;
http2 on;
server_name auth.${BASE_DOMAIN};
ssl_certificate ${APP_DIR}/certs/auth.${BASE_DOMAIN}/cert.pem;
ssl_certificate_key ${APP_DIR}/certs/auth.${BASE_DOMAIN}/key.pem;
client_max_body_size 4m; # skin uploads
location / {
proxy_pass http://127.0.0.1:25585;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto $scheme;
}
}

86
plan/14-deploy.md Normal file
View File

@@ -0,0 +1,86 @@
# Deployment — redeploy to `cochi`
Production host for the Ulicraft stack.
| | |
|---|---|
| Host | `cochi` (SSH alias) |
| Path | `/home/ubuntu/mc/ulicraft-server-v1` |
| Branch | `main` |
| Run mode | `online` (compose files: `docker-compose.yml` + `docker-compose.static.yml`) |
| Auth | Drasl (base stack — no Blessing Skin override) |
| Restart | **hard** (`compose down``build-stack.sh --up`) — brief full downtime |
## Routine redeploy
```bash
ssh cochi
cd /home/ubuntu/mc/ulicraft-server-v1
git pull --ff-only
# hard restart: tear the stack down, then bring it back up (renders configs first)
docker compose -f docker-compose.yml -f docker-compose.static.yml down --remove-orphans
tooling/build-stack.sh --up online
```
Or as a one-shot from your workstation:
```bash
ssh cochi 'set -e
cd /home/ubuntu/mc/ulicraft-server-v1
git pull --ff-only
docker compose -f docker-compose.yml -f docker-compose.static.yml down --remove-orphans
tooling/build-stack.sh --up online'
```
`build-stack.sh --up online` runs preflight (checks `.env`), `render-config.sh`
(which renders configs + `render-pack.sh` for the packwiz pack), then
`docker compose ... up -d` with the online file set.
## What survives a hard restart
Named volumes persist — world and auth data are safe:
- `mc_data` — world + installed NeoForge/mods
- `drasl_state` — Drasl accounts/skins
- `backups/` — mc-backup snapshots
`down` removes **containers**, not volumes. Players are disconnected during the
restart (a few minutes); they reconnect once `minecraft` is healthy again.
## Prerequisites (one-time, on `cochi`)
- SSH access as the `cochi` alias.
- `.env` present and complete: `BASE_DOMAIN`, `HOST_LAN_IP`, `RCON_PASSWORD`
(`render-config.sh` halts on any unset var, even in online mode).
- Docker + compose plugin, `packwiz`, `envsubst`, `git`.
- **First deploy only** — heavy online prep (fetch launcher, pre-bake the server
volume, etc.):
```bash
tooling/build-stack.sh --prep
```
Not needed for routine redeploys; `--up` alone re-syncs the pack and configs.
## Verify
```bash
docker compose -f docker-compose.yml -f docker-compose.static.yml ps
docker compose -f docker-compose.yml -f docker-compose.static.yml logs -f minecraft # watch for "Done"
curl -fsS "http://auth.$(grep '^BASE_DOMAIN=' .env | cut -d= -f2)/" >/dev/null && echo "auth ok"
```
## Rollback
```bash
git log --oneline -5 # find the last-good commit
git checkout <sha> # or: git reset --hard <sha>
docker compose -f docker-compose.yml -f docker-compose.static.yml down --remove-orphans
tooling/build-stack.sh --up online
```
## Notes
- **Online mode needs internet** on `cochi` at restart (no air-gap mirror): the
server installs NeoForge + mods over the network. For an internet-less deploy,
switch to `airgap` (needs `--prep` first) — see `plan/13-dns-and-run-modes.md`.
- Switching to the **Blessing Skin** auth variant changes the file set to
`-f docker-compose.yml -f docker-compose.blessingskin.yml` (see
`plan/03b-blessing-skin.md`); update the `down`/`up` commands accordingly.

76
plan/15-letsencrypt.md Normal file
View File

@@ -0,0 +1,76 @@
# TLS — Let's Encrypt (OVH DNS-01) + host nginx ingress
Production serves real, publicly-trusted certs via the **host's nginx**. Caddy is
no longer the production ingress — it moved to its own compose file for the
LAN/air-gap path.
## Two ingress paths
| Path | Ingress | Compose | TLS |
|------|---------|---------|-----|
| Production | host **nginx** | `docker-compose.yml` + `docker-compose.nginx.yml` | Let's Encrypt (this doc) |
| LAN / air-gap | **caddy** container | `docker-compose.yml` + `docker-compose.caddy.yml` (+ static/mirror) | Caddy internal CA |
`build-stack.sh` drives the caddy path (adds `docker-compose.caddy.yml`
automatically). The nginx path is a manual deploy (below). This supersedes the
ingress part of `plan/14-deploy.md` for production.
## Certs: one per name, via OVH DNS-01
`tooling/issue-letsencrypt.sh` issues a **separate** cert for the apex and each
subdomain (`${BASE_DOMAIN}`, `auth.…`, `pack.…`) using acme.sh + the OVH DNS-01
challenge — no inbound ports needed, so it works even on a LAN-only host.
### One-time setup
1. Install acme.sh:
```bash
curl https://get.acme.sh | sh -s email=you@example.com
```
2. OVH API credentials → `.env` (see `.env.example` `OVH_*`, `LE_EMAIL`,
`LE_SUBDOMAINS`). Create a token at <https://eu.api.ovh.com/createToken/> with
`GET/POST/PUT/DELETE` on `/domain/zone/*`, or leave `OVH_CK` blank and run the
script once — acme.sh prints an authorization URL; visit it, then paste the
consumer key into `OVH_CK` and re-run.
### Issue
```bash
tooling/issue-letsencrypt.sh # real certs
LE_STAGING=1 tooling/issue-letsencrypt.sh # dry run against the LE staging CA
```
Certs install to `certs/<name>/{cert.pem,key.pem}` (gitignored). acme.sh adds a
cron for renewal; `--reloadcmd` reloads nginx automatically on renew. Override
the reload with `RELOAD_CMD=...` if nginx isn't systemd-managed.
## nginx
1. Stack up with the nginx override (publishes drasl on `127.0.0.1:25585`):
```bash
docker compose -f docker-compose.yml -f docker-compose.nginx.yml up -d
```
2. Render + install the vhost (`nginx/ulicraft.conf.tmpl`):
```bash
APP_DIR=/home/ubuntu/mc/ulicraft-server-v1 BASE_DOMAIN=ulicraft.net \
envsubst '$APP_DIR $BASE_DOMAIN' \
< nginx/ulicraft.conf.tmpl | sudo tee /etc/nginx/sites-available/ulicraft.conf
sudo ln -sf /etc/nginx/sites-available/ulicraft.conf /etc/nginx/sites-enabled/
sudo nginx -t && sudo systemctl reload nginx
```
nginx then: apex + `pack.*` as static files (`www`, `pack`, `custom`),
`auth.*` → drasl. Ensure `www-data` can read `$APP_DIR/{www,pack,custom,certs}`.
## DNS
Point the public/party DNS for `${BASE_DOMAIN}`, `auth.${BASE_DOMAIN}`,
`pack.${BASE_DOMAIN}` at the host. The Minecraft container reaches them via
`extra_hosts: host-gateway` (set in `docker-compose.nginx.yml`) and uses HTTPS —
the JVM trusts Let's Encrypt with no CA import (the big win over the internal CA).
## Why DNS-01 (not HTTP-01)
`HOST_LAN_IP` shows a LAN host; HTTP-01/TLS-ALPN-01 need public inbound on 80/443.
DNS-01 only needs the OVH API, so it issues regardless of public reachability —
and `ulicraft.net` is a real OVH-managed domain, which LE requires (it never
issues for `.lan`/`.local`).

View File

@@ -91,7 +91,8 @@ prep() {
log "pre-bake server volume (TYPE=NEOFORGE install, one-shot, online)"
# Bring minecraft up once so itzg installs NeoForge + mods into the volume,
# then stop it. The party then runs offline against the baked volume.
"${COMPOSE[@]}" up -d minecraft
# caddy.yml so the pack (pack.${BASE_DOMAIN}) resolves + is served during bake.
"${COMPOSE[@]}" -f docker-compose.yml -f docker-compose.caddy.yml up -d minecraft
warn "watch logs until 'Done' then re-run with --up; or automate a readiness probe"
}
@@ -102,7 +103,10 @@ prep() {
# core base only (no landing, no mirror) — debugging
up() {
local mode="${1:-airgap}"
local -a files=(-f docker-compose.yml)
# build-stack is the caddy (LAN/air-gap) deployer; caddy lives in its own file
# now (the base stack can also run behind the host's nginx — see
# docker-compose.nginx.yml). Always include caddy here.
local -a files=(-f docker-compose.yml -f docker-compose.caddy.yml)
case "$mode" in
core) ;;
online) files+=(-f docker-compose.static.yml) ;;

101
tooling/issue-letsencrypt.sh Executable file
View File

@@ -0,0 +1,101 @@
#!/usr/bin/env bash
# issue-letsencrypt.sh — issue per-name Let's Encrypt certificates via the OVH
# DNS-01 challenge (acme.sh) and install them where Caddy serves them.
#
# Run MANUALLY on the host. One separate certificate per name:
# ${BASE_DOMAIN} + <sub>.${BASE_DOMAIN} for each sub in LE_SUBDOMAINS.
# Certs land in certs/<name>/{cert.pem,key.pem} (gitignored) and the host nginx
# is reloaded so it picks them up (see nginx/ulicraft.conf.tmpl).
#
# DNS-01 needs no inbound ports — works for a LAN-only host. It only needs OVH
# API credentials with DNS-record write access for the zone.
#
# .env vars required:
# BASE_DOMAIN e.g. ulicraft.net
# LE_EMAIL account email for Let's Encrypt
# OVH_AK OVH_AS OVH application key + secret
# OVH_CK OVH consumer key (obtained on first run — see below)
# OVH_END_POINT OVH API endpoint (e.g. ovh-eu)
# optional:
# LE_SUBDOMAINS space-separated (default: "auth pack")
# LE_STAGING=1 use the LE staging CA (untrusted, for dry runs)
#
# First-time OVH consumer key: set OVH_AK/OVH_AS/OVH_END_POINT, leave OVH_CK
# empty, run this once — acme.sh prints an authorization URL. Visit it, grant
# access, then copy the consumer key it shows into OVH_CK in .env and re-run.
# (Or create a token directly: https://eu.api.ovh.com/createToken/ with
# GET/POST/PUT/DELETE on /domain/zone/*.)
set -euo pipefail
cd "$(dirname "$0")/.." # repo root
log() { printf '\n\033[1;32m==>\033[0m %s\n' "$*"; }
warn() { printf '\033[1;33m!!\033[0m %s\n' "$*" >&2; }
die() { printf '\033[1;31mxx\033[0m %s\n' "$*" >&2; exit 1; }
# ---- config / preflight ----------------------------------------------------
[ -f .env ] || die ".env not found (copy .env.example and fill it)"
# shellcheck disable=SC1091
set -a; . ./.env; set +a
: "${BASE_DOMAIN:?BASE_DOMAIN unset in .env}"
: "${LE_EMAIL:?LE_EMAIL unset in .env}"
: "${OVH_AK:?OVH_AK unset in .env}"
: "${OVH_AS:?OVH_AS unset in .env}"
: "${OVH_END_POINT:?OVH_END_POINT unset in .env (e.g. ovh-eu)}"
# OVH_CK may be empty on the very first run (acme.sh will print the auth URL).
export OVH_AK OVH_AS OVH_END_POINT
export OVH_CK="${OVH_CK:-}"
ACME="${ACME_SH:-$HOME/.acme.sh/acme.sh}"
[ -x "$ACME" ] || ACME="$(command -v acme.sh || true)"
[ -n "$ACME" ] && [ -x "$ACME" ] || die "acme.sh not found. Install:
curl https://get.acme.sh | sh -s email=${LE_EMAIL}
then re-run (or set ACME_SH=/path/to/acme.sh)."
CERT_DIR="${CERT_DIR:-certs}"
read -r -a SUBDOMAINS <<< "${LE_SUBDOMAINS:-auth pack}"
if [ "${LE_STAGING:-0}" = "1" ]; then
SERVER="letsencrypt_test"; warn "STAGING mode — certs will NOT be trusted by browsers"
else
SERVER="letsencrypt"
fi
# Reload the host nginx after each install so new certs take effect. Override
# RELOAD_CMD if nginx runs elsewhere (e.g. a container or a non-systemd host).
RELOAD="${RELOAD_CMD:-sudo systemctl reload nginx} || true"
# Full name list: apex + each subdomain.
domains=("$BASE_DOMAIN")
for s in "${SUBDOMAINS[@]}"; do domains+=("${s}.${BASE_DOMAIN}"); done
log "CA=$SERVER names: ${domains[*]}"
# Register the ACME account once (idempotent).
"$ACME" --register-account -m "$LE_EMAIL" --server "$SERVER" >/dev/null 2>&1 || true
# ---- issue + install one cert per name --------------------------------------
issued=0
for d in "${domains[@]}"; do
log "issuing $d (OVH DNS-01)"
rc=0
"$ACME" --issue --dns dns_ovh --server "$SERVER" --keylength ec-256 -d "$d" || rc=$?
# acme.sh: 0 = issued/renewed, 2 = skipped (cert still valid, not due). Both ok.
case "$rc" in
0|2) ;;
*) die "acme.sh failed for $d (rc=$rc). If this is the first OVH run, follow
the consumer-key authorization URL printed above, set OVH_CK in .env, re-run." ;;
esac
mkdir -p "$CERT_DIR/$d"
"$ACME" --install-cert -d "$d" --ecc \
--key-file "$CERT_DIR/$d/key.pem" \
--fullchain-file "$CERT_DIR/$d/cert.pem" \
--reloadcmd "$RELOAD"
echo " installed -> $CERT_DIR/$d/{cert.pem,key.pem}"
issued=$((issued+1))
done
log "done — $issued certificate(s) under $CERT_DIR/"
echo " serve them: point nginx at certs/<name>/ — see nginx/ulicraft.conf.tmpl"
echo " renewals: acme.sh installs a cron; --reloadcmd reloads nginx automatically."