* Fix TLS handshake failure for any host without a cached target Reported: fallback SSL doesn't work in the Docker build. Reproduced — it's worse than the fallback specifically: TLS was broken for nearly every connection, including ones with no SNI at all: $ curl -vk https://127.0.0.1/ * TLSv1.3 (IN), TLS alert, internal error (592) * OpenSSL/3.0.13: error:0A000438:SSL routines::tlsv1 alert internal error Root cause: targetinfo.lua's M.get() is shared by two call sites in two incompatible nginx phases — - proxy.conf's access_by_lua_block (a normal HTTP request phase, where ngx.exit() is valid) - nginx.conf's request_domain callback, which runs during the TLS handshake itself (ssl_certificate_by_lua*), where ngx.exit() is NOT a supported API M.get() called ngx.exit() on every lookup failure (no domain/SNI, a Redis error, or an unregistered host). When invoked from the SSL phase, that aborted the handshake with a bare "internal error" alert and produced no log output anywhere — silent and total, not limited to the unregistered-domain case, since even a connection with no SNI hits the same code path immediately. Fix: M.get() no longer calls ngx.exit() itself — it returns (nil, httpStatus) on failure. proxy.conf now checks the return value and calls ngx.exit() itself (the phase where that's actually supported). nginx.conf's request_domain guards the now-possibly-nil result before indexing it, and leaves ngx.ctx.toAllow unset on failure so allow_domain() correctly denies issuance and auto-ssl falls through to the static fallback cert in autossl.conf. Verified end to end against a running Docker build (deployed the changed files into a live container and reloaded, rather than relying on a full rebuild each iteration): - No SNI at all: TLS now completes; HTTP layer correctly returns 406 (previously: broken handshake, no response at all) - Unregistered SNI: same — TLS completes, 406, and openssl s_client confirms the cert served is genuinely the fallback (CN=sni-support-required-for-valid-ssl) - A real registered Host: TLS completes and proxies through to the backend correctly (confirms the success path is unaffected) - npm test: 192/192 pass * Fix footer not sticking to the bottom on short pages body had no sticky-footer layout at all (sso-manager-node already had this; proxy never did), so on any page with little content (e.g. /login) the footer sat right after the content instead of at the bottom of the viewport, leaving a large gap below it. Added the same flex-based pattern already used in sso-manager-node: body is a column flex container, #spa-shell grows to fill the remaining space, pushing the footer (the next sibling) to the bottom. Verified visually (screenshot) and via computed layout (footer.getBoundingClientRect().bottom === window.innerHeight) before and after. * Fix commit hash not showing in Docker builds build_info.js computed buildHash via `git rev-parse --short HEAD` at runtime, but the final image intentionally has no git binary and no .git directory (kept lean, per .dockerignore) — so this always failed silently and the footer's version line showed "unknown" for every Docker deployment. Working correctly only for bare-metal/dev, where git + .git are actually present. Added a throwaway gitinfo build stage that reuses the main base image (no extra pull) with git installed just for this stage, reads .git from the build context (now no longer excluded — see .dockerignore), and bakes the resolved short hash into a small file that IS copied into the final image. build_info.js reads that file first, falling back to the old git-rev-parse behavior (still needed for bare-metal). Verified against a real build: `docker exec proxy cat /app/.build_commit` matches `git rev-parse --short HEAD` on the host, and the footer now shows the real hash instead of "unknown". * Allow the local anti-lockout admin's initial password to be configured The local "proxyadmin2" bootstrap account was always created with username == password == "proxyadmin2" — a hardcoded, publicly-known default with no way to set it to something else before first boot. Fine for a quick local test, not for anything exposed publicly, and orchestrators like theta-env's setup.sh (which already generates a random password for the SSO admin) had no way to do the same here. Added conf.auth.localAdminPass (proxy-secrets.js / app_auth__localAdminPass): if set, it's used as the initial password instead of the hardcoded default. Only read on first creation — once the account exists this is never consulted again, so it's safe to leave set. Falls back to the previous behavior (password == username) when unset, so this is fully backward compatible. Verified: with app_auth__localAdminPass set, login with the new password succeeds and the old default ("proxyadmin2") is correctly rejected. Confirmed in a real Docker build too (secrets.js auth.localAdminPass), and npm test 192/192 pass. * Support GIT_COMMIT build-arg override for submodule builds The gitinfo stage from the previous commit works for a standalone clone (.git is a real directory) but not when this repo is built as a git submodule (e.g. from theta-env): a submodule's .git is a pointer FILE, not a directory — the real object database lives in the superproject's .git/modules/, outside this repo's own directory and therefore outside Docker's build context entirely. `git rev-parse` can never resolve it from in here no matter what, so builds via theta-env still baked in "unknown" despite the earlier fix. Add an optional GIT_COMMIT build-arg that, when set, wins over the in-context git resolution. theta-env's setup.sh now computes it on the host (where the submodule DOES resolve correctly) and passes it via docker-compose.yml's build.args. Verified via theta-env's actual setup.sh end to end: rebuilding with this change, `docker exec proxy cat /app/.build_commit` now matches `git -C proxy rev-parse --short HEAD` on the host (previously: "unknown", confirmed via the "[Warning] One or more build-args [GIT_COMMIT] were not consumed" message before this fix synced into the docker-compose.yml side).
12 KiB
Deployment Guide — theta42/proxy
The proxy is an OpenID Connect-protected reverse proxy (OpenResty front + Node management app + Redis) that is both an OIDC client of an SSO Manager and a direct LDAP client for user lookups. Two deployment methods:
- Docker — a single all-in-one image bundling OpenResty + the app + Redis
(
docker compose up). - Bare metal —
ops/install.shon Debian/Ubuntu (installs Node.js, OpenResty, Lua modules, Redis, and a systemd unit).
How configuration works
The app loads configuration via @simpleworkjs/conf,
which deep-merges, in order:
conf/base.js(committed, generic defaults)conf/<NODE_ENV>.js(optional)conf/secrets.js(gitignored — secrets + per-deployment values)app_*environment variables — the highest-precedence layer
Any env var whose name starts with app_ overrides the merged config. The rest
of the name is split on double-underscore (__) into a nested path. Values
are JSON.parse-coerced when possible and kept as raw strings otherwise.
| Env var | Sets | Notes |
|---|---|---|
app_oidc__issuer |
conf.oidc.issuer |
browser-facing SSO URL |
app_oidc__authorizationEndpoint |
conf.oidc.authorizationEndpoint |
browser-facing |
app_oidc__tokenEndpoint |
conf.oidc.tokenEndpoint |
server-to-server; can be internal |
app_oidc__userinfoEndpoint |
conf.oidc.userinfoEndpoint |
server-to-server; can be internal |
app_oidc__endSessionEndpoint |
conf.oidc.endSessionEndpoint |
browser-facing |
app_oidc__clientId / app_oidc__clientSecret |
OIDC client creds | register in the SSO first |
app_oidc__redirectUri |
conf.oidc.redirectUri |
must match the SSO client exactly |
app_oidc__enabled |
conf.oidc.enabled |
boolean |
app_ldap__url |
conf.ldap.url |
ldaps://…:636 or ldap://…:389 |
app_ldap__bindDN / app_ldap__bindPassword |
LDAP service account | don't reuse the admin DN |
app_ldap__searchBase / app_ldap__userFilter |
user search | |
app_ldap__tlsOptions__rejectUnauthorized |
conf.ldap.tlsOptions.rejectUnauthorized |
false for self-signed LDAPS |
app_ldap__tlsOptions__ca |
conf.ldap.tlsOptions.ca |
path to a CA cert for strict trust |
app_auth__adminUsers |
conf.auth.adminUsers |
local anti-lockout admin (uid) |
app_auth__adminGroups |
conf.auth.adminGroups |
SSO/LDAP groups that are global admin (JSON array) |
app_auth__localAdminPass |
conf.auth.localAdminPass |
initial password for the local anti-lockout admin (used once, on first creation only — defaults to the username itself if unset) |
app_redis__prefix |
conf.redis.prefix |
default proxy_ |
See docs/docker.md for a shorter, container-focused version
of this reference.
Requires
@simpleworkjs/conf>= 1.1.0. The Docker image will not honorapp_*env vars on 1.0.0. The lock is already on^1.1.0; if you regenerate it:cd nodejs && npm install @simpleworkjs/conf@^1.1.0
Method 1: Docker (all-in-one)
The image (Dockerfile) bundles OpenResty + the app + Redis in one container,
mirroring the bare-metal ops/install.sh layout. docker-entrypoint.sh:
generates the fallback SSL cert, parameterizes the OpenResty resolver/
set_real_ip_from directives, starts Redis + the node app, and execs OpenResty
in the foreground under dumb-init.
Setup
The bundled docker-compose.yml reads the OIDC + LDAP + auth wiring from a
bind-mounted ./config/proxy-secrets.js (not from app_* env). Copy the
example, fill in your secrets, then build + start:
mkdir -p config && chmod 700 config
cp secrets.js.example config/proxy-secrets.js
$EDITOR config/proxy-secrets.js # set oidc.clientId/clientSecret, ldap.bindPassword, ...
docker compose up -d --build
docker-entrypoint.sh symlinks /config/proxy-secrets.js → /app/conf/secrets.js
so @simpleworkjs/conf reads it. No app_* env is passed — app_* env would
override the file (env beats secrets.js in @simpleworkjs/conf), so the file is
kept authoritative. RESOLVER / REAL_IP_FROM / NODE_ENV / NODE_PORT are
OpenResty-runtime / process env, not app_* config, so they stay in the compose.
Running the unified
theta-envstack? Itssetup.shgenerates./config/proxy-secrets.js(+./config/sso-secrets.js) for you and registers the OAuth client with the SSO — see the theta-env README.
Access
- Proxy (public):
https://<host>/— OpenResty front, auto-SSL (Let's Encrypt) - Management UI / API:
http://127.0.0.1:3000/(bound to localhost; the front proxies the UI under its own TLS) - Health:
http://127.0.0.1:3000/health→{"status":"ok"}
API tokens (personal access tokens)
Any logged-in user can mint a long-lived bearer token to call the management
API from scripts/CI/other services, without an OIDC browser session. Tokens are
self-service and authenticate as their creator: the creator's groups are
snapshotted at mint time (mirroring how the proxy's browser session captures
groups at login — the proxy never re-queries the IdP), and the existing authz
layer (Permission.effectiveFor / roles.resolveEffective) applies unchanged.
Local groups and owned-domain rights are recomputed live each request; only the
SSO/LDAP group membership is the mint-time snapshot.
Create one in the UI under API Tokens (the token string is shown once), then use it as a bearer token:
curl -H "Authorization: Bearer prx_<id>_<secret>" https://proxy.example.com/api/host
Format: prx_<id>_<secret> — the id is the lookup key, the secret is
bcrypt-hashed and never stored in plaintext. Rotate or revoke from the same page
(immediate effect). Optional expiry (in days) at creation. Tokens persist in the
bundled Redis (AOF — see Backups and restore), so they survive rebuilds.
The token carries the creator's effective rights: a global admin's token can
manage Hosts/Users/Groups; a domain manager's token can manage their own
domains but requireAdmin routes return 403. To tighten permissions after group
changes, revoke and re-mint the token.
OpenResty runtime env
| Variable | Default | Description |
|---|---|---|
RESOLVER |
127.0.0.11 |
DNS for upstream names in Host records (Docker's embedded DNS) |
REAL_IP_FROM |
empty | Trusted CIDR for X-Real-IP. Empty = proxy is the front (removes the real_ip block). Set to an upstream proxy's CIDR if one sits in front. |
Auto-SSL / Let's Encrypt
lua-resty-auto-ssl stores certs in the bundled Redis. Redis is now AOF+RDB
persisted to the proxy-data volume (not in-memory), so Let's Encrypt certs
survive container recreation — no re-issue / rate-limit on rebuild. Port 80 is
required for HTTP-01 challenges (mapped in the compose).
Backups and restore
What lives where
| State | Location | Persisted? |
|---|---|---|
| Host records, permissions, DNS creds, local users | proxy-data volume (/data, Redis) |
yes (AOF + RDB) |
| Let's Encrypt certs (auto-ssl) | proxy-data volume (/data, Redis) |
yes — same Redis |
| nginx response cache / logs | proxy-cache / proxy-logs volumes |
yes (volume) |
| Secrets (OIDC client secret, LDAP bind password) | ./config/proxy-secrets.js (bind mount) |
your responsibility — back up off-host |
Automatic snapshots — when run as part of the unified theta-env stack,
setup.sh snapshots Redis + ./config/ to ./backups/<timestamp>/ before every
rebuild and keeps the last BACKUP_KEEP (default 5). Standalone deployments use
the manual steps below.
Manual backup
# Redis — hot snapshot: trigger a save, then copy the RDB out
docker compose exec proxy redis-cli BGSAVE
docker compose cp proxy:/data/dump.rdb proxy-redis-$(date +%F).rdb
# Secrets — copy the config dir (holds OIDC client secret, LDAP bind password)
cp -a ./config config-backup-$(date +%F) && chmod 700 config-backup-$(date +%F)
Store the .rdb and config copy off the host — they contain secrets and
the whole Host/permission/user dataset.
Restore — Redis (full proxy state + certs)
cp -a config-backup-<date> ./config && chmod 700 ./config
docker compose up -d
docker compose stop proxy
# AOF wins on startup — delete it so the RDB is loaded instead (see note).
docker compose run --rm --no-deps --entrypoint sh proxy -c \
'rm -f /data/appendonly.aof /data/appendonly.aof.*'
docker compose cp proxy-redis-<date>.rdb proxy:/data/dump.rdb
docker compose start proxy
AOF vs RDB (important): with
--appendonly yes, Redis loadsappendonly.aofon startup and ignoresdump.rdbif the AOF exists. To restore from an RDB snapshot you must delete the AOF first (the runbook does this); Redis then loads the RDB and writes a fresh AOF. Verify:docker compose exec proxy redis-cli DBSIZE.Restoring Redis restores cert state at the snapshot time — certs issued after the snapshot are lost and will be re-issued on next request.
Upgrades
./setup.sh # backs up, then rebuilds — proxy-data keeps Redis state
# (standalone) docker compose pull && docker compose up -d
Host records, permissions, DNS creds, local users, and Let's Encrypt certs all
survive the rebuild because they live on the proxy-data volume, not in the
image. Migrations note: if a release ships a nodejs/migrations/ script,
run it after upgrading (it transforms in-Redis records); see the release notes.
Logs
OpenResty runs in the foreground and the Node app in the background, both
writing to the container's stdout/stderr. nginx access/error logs go to files
(/var/log/nginx, on the proxy-logs volume), so they do not appear in
docker logs.
docker compose logs -f proxy # app + OpenResty (stdout/stderr)
docker compose exec proxy tail -f /var/log/nginx/error.log # nginx errors
docker compose exec proxy tail -f /var/log/nginx/access.log # nginx access
docker compose logs --tail=200 --since=10m proxy # recent context
Method 2: Bare metal (Debian/Ubuntu)
ops/install.sh is an idempotent installer: it installs Node.js 22.x, OpenResty
(from openresty.org), Lua modules (luarocks), Redis, force-syncs the repo to
/var/www/proxy, symlinks the OpenResty + systemd config from the repo, and
starts proxy.service. Re-run it to update.
sudo ./ops/install.sh
Configuration is file-based: write nodejs/conf/secrets.js with the OIDC +
LDAP values (see nodejs/conf/base.js for the shape), then
sudo systemctl restart proxy.
Fronting an SSO Manager
The proxy is a natural front for theta42/sso-manager-node:
it terminates TLS for the SSO's UI and protects it with OIDC login, while also
binding to the SSO's LDAP directly for user lookups. To run both together:
- One Docker network so the proxy can reach the SSO internally at
http://sso-manager:3001(token/userinfo, server-to-server) andldaps://sso-manager:636(LDAP). - Set the SSO's
OAUTH_ISSUERto the browser-facing HTTPS URL the proxy serves the SSO at (e.g.https://sso.example.com). - Register the proxy as an OIDC client in the SSO, with a
redirectUrimatching the proxy's callback (https://proxy.example.com/api/auth/oidc/callback). - LDAP: point
app_ldap__urlatldaps://sso-manager:636and create a dedicated service account (cn=ldapclient,ou=people,…) — don't reuse the admin DN. For the SSO's self-signed LDAPS cert, setapp_ldap__tlsOptions__rejectUnauthorized=false(or mount the cert and useapp_ldap__tlsOptions__ca=<path>).
The theta42/theta-env unified repo
automates all four steps with ./setup.sh.
Security notes
- Never commit
secrets.js— it's in.gitignore. - Bind the management port to localhost (the compose does:
127.0.0.1:3000). The OpenResty front proxies the UI under TLS; don't expose 3000 to the LAN. REAL_IP_FROMempty by default — the proxy trusts no one to setX-Real-IP(it's the front). Only set it if a trusted proxy sits in front.- LDAPS for any LDAP that crosses the network. Use
ldaps:///StartTLS; plainldap://is fine only on a private docker network. - The image runs OpenResty workers as
nobodyand the node app as root (matches the bare-metal systemd unit). Harden to a non-root user for production if needed.