The OIDC provider sends an unauthenticated authorize request through
/login/oauth/authorize?client_id=…&state=…; dropping the query there
loses the whole authorization request. The ?redirect= form is unaffected.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
formAJAX always passes the serialized form as the second argument, so a
DELETE-method form (the host/DNS delete buttons) landed its callback in
the data slot and never ran.
The login page's "reveal the card once we know you're logged out" branch
touched an element further down the same page, which threw when
isLoggedIn answered before the parser got there (it always did without a
stored token). It now runs on DOM ready.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
views/top.ejs, views/bottom.ejs and public/lib/js/app-base.js are now
byte-identical across sso-manager-node, proxy and jump-host. Everything
per-app moved into utils/ui.js, exposed to every render as `ui` via
app.locals (nav items + their group gates, footer repo/docs/ToS links,
favicon, profile/logout targets, update-banner on/off + label).
Client framework changes:
- One gating model everywhere: app-base.js reveals .group-required-<cn>
for each of the current user user/me groups. sso-manager-node sends LDAP
DNs in memberOf, the OIDC clients send CNs in groups; both normalise to
CNs, and the clients isAdmin flag becomes a synthetic `admin` group, so
proxy nav-admin items are now group-required-admin.
- user/me is fetched once per page load and cached (app.auth.loadUser);
nav, forceLogin and group-required elements all read that one promise.
- isLoggedIn is dual-mode (Promise + node-style callback), so the async
and callback call styles both work from one shared top.ejs.
- forceLogin no longer uses $.holdReady (removed in jQuery 4): it redirects
to /login?redirect=<path>, and still enforces required groups.
- logOut only clears the session; the caller decides where to go next.
- post/put/delete are dual-mode Promise/callback, which also removes the
undefined `callback2` reference that threw on a non-function callback.
Dependencies: jquery ^4.0.0 and ejs ^3.1.10 in all three apps.
proxy specifics:
- .group-required base rule added to styles.css; the admin nav items lost
their inline display:none in favour of it.
- The brand link points at / instead of #.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Rewire onto the shared @simpleworkjs/oidc-client, /ldap, and /app-stack
packages (deleting the byte-identical local forks of the same code), close the
LDAP filter-injection in User.get by routing the username through escapeFilter
(RFC 4515), align model-redis ^1.6.0 and ldapts ^8.1.8, and unify build_info to
{buildVersion, buildHash, buildYear}. package-lock regenerated from the npm
registry (no file:/link:), so npm ci is clean in docker builds.
Co-Authored-By: Claude <noreply@anthropic.com>
Every request to a host with additional load-balancing targets 500'd:
targetinfo.lua required 'resty.balancer.round_robin', which does not
exist in the lua-resty-balancer rock actually installed by the
Dockerfile/install.sh. That rock provides resty.roundrobin instead,
with a different constructor (roundrobin:new(nodes), not
:new() + :reinit(nodes)).
Verified end-to-end in a rebuilt image: requests to a load-balanced
host now return 200 and alternate across both backend targets, with
no Lua errors in the OpenResty log.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The load-balancing feature (merged in #172/#173) was tagged v1.2.0 on
GitHub, but nodejs/package.json and CHANGELOG.md were never bumped for
it. Backfill a 1.2.0 CHANGELOG entry and bump this release to 1.2.1
instead of 1.1.18, so package.json/CHANGELOG stay ahead of the highest
existing tag rather than behind it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
models/user_redis.js hardcoded the bootstrap anti-lockout admin
username to 'proxyadmin2', while migrations/permission_bootstrap.js
grants the global-admin permission to conf.auth.adminUsers[0]. An
operator who customized adminUsers away from the default ended up
with a bootstrapped account that had no admin permissions -- a
silent lockout. user_redis.js now derives the bootstrap username the
same way permission_bootstrap.js does.
Also corrected a secrets.js.example comment that claimed the
bootstrap password defaults to the username itself (it actually
generates and logs a random password), and refreshed all README
screenshots against the current UI, including a new load-balancing
screenshot.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Fixes#47.
- Added lua-resty-balancer to dependencies (Dockerfile & install.sh).
- Added 'targets' field to the Host model to hold additional targets.
- Updated the UI to allow inputting additional targets (IP:port).
- Updated targetinfo.lua to parse the additional targets and load balance between them and the primary target using resty.balancer.round_robin.
Move the wildcard sibling-parent fix into its own 1.1.17 section (it is not
in the v1.1.16 tag) and bump nodejs/package.json + lockfile to 1.1.17.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Host.lookUpWildcardParent() walked all labels of the host down to its own
leaf and only inspected that leaf's "*" child, so it found a wildcard
nested under the host (the base-domain case, e.g. *.cool.mysite.com for
cool.mysite.com) but missed the common case where the wildcard is a
SIBLING of the host's leftmost label (e.g. *.nl.wgnode.com covering an
already-existing sso.nl.wgnode.com). The /wildcard-parent route then
returned nothing and the edit form's "Parent Wildcard" option stayed
greyed out, leaving no way to convert an existing auto-SSL host onto a
wildcard issued afterward.
Track the parent node during the walk and check the sibling "*" slot too.
The never-created-subdomain case is unchanged (plain lookUp()'s wildcard
fallback in the route still handles it). Sync the test mock and add
regression coverage for the sibling case (existing host, never-created
host via the route fallback, and a deeper-wildcard negative case).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Replace isomorphic-dompurify with xss to avoid ESM-only jsdom/htmlparser2
transitive deps that break the existing Jest test suite.
- Keep rendered docs HTML sanitized via xss() in routes/docs.js.
Co-Authored-By: Claude <noreply@anthropic.com>
- Add isomorphic-dompurify to sanitize rendered docs HTML
- Tighten SocketServerJson socket permissions from 777 to 660
- Keep package.json version at 1.1.16
Co-Authored-By: Claude <noreply@anthropic.com>
- Fix MIT LICENSE copyright placeholder
- Remove private flag and correct GitHub repository URL in package.json
- Bump version to 1.1.16
- Genericize committed config defaults (example.com/localhost)
- Harden global error handler against information leakage
- Generate random initial password for proxyadmin2 bootstrap account
- Correct docs to describe CONF_SECRETS instead of symlink behavior
Co-Authored-By: Claude <noreply@anthropic.com>
conf 1.2.0 adds CONF_SECRETS, an env var to point at the secrets file
directly -- use it in the Docker entrypoint instead of symlinking the
mounted file into /app/conf/secrets.js, so the app no longer needs
write access to its own conf/ directory to pick up mounted secrets.
jq-repeat 2.2.0 is a compatible feature release (sort(), replace(),
faster leading-edge update() timing); no call-site changes needed.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- New docs/concepts-{hosts,dns,access,api-tokens}.md -- plain-language
guides aimed at less technical readers, each linking onward to the
existing system-design-level doc for anyone who wants that detail.
Card help links (Proxy List, Add/Edit host, DNS Provider cards,
Users/Permissions/Groups cards) now point here instead of straight at
Installation/Architecture.
- The "New API Token" card had no help link at all -- added, pointing to
the new API Tokens doc.
- Fixed the in-app docs viewer rendering every docs/*.md page with a
garbled heading + stray <hr> at the top: Jekyll front matter (meant
only for the GitHub Pages build) was never stripped before being
handed to the markdown renderer.
- Fixed cross-doc links never resolving in-app, since this viewer serves
docs at /docs/<slug> with no .html suffix: rewritten to the correct
in-app URL, first by registered slug, falling back to the doc's real
filename (the correct, working link form on the Jekyll/GitHub Pages
build) -- same idea as the existing image-path fix, and lets one link
written in a doc work on both targets.
Bumps to v1.1.13.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KDEx8ghuZR61pqPXc6da9C
- The edit form's "Parent Wildcard" option stayed greyed out even when a
valid wildcard existed, since hostEditOpen() never ran the eligibility
check (only the host field's keyup handler did, which setting .val()
programmatically doesn't fire) -- and the check itself, GET
/host/lookup/:item, had the same self-match bug as the recently-fixed
Host.prototype.update() case: it resolves an already-existing host to
its own record instead of a sibling wildcard. Added a dedicated
/host/wildcard-parent/:item route combining lookUp() (handles a
brand-new subdomain) with lookUpWildcardParent() (handles an
already-existing host), and hostEditOpen() now actually runs it.
- Migrated ops/nginx_conf/autossl.conf's deprecated "listen ... http2"
directive to the standalone "http2 on;" directive (nginx 1.25.1+).
Bumps to v1.1.12.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KDEx8ghuZR61pqPXc6da9C
The single header-wide help icon (added last release) pointed at a
per-page doc guess, but a page can have several cards covering different
topics. Removed it and added a small help icon directly to each card
that has real corresponding doc content, linking straight to that doc:
Proxy List + Add/Edit host modal (hosts.ejs), Add DNS Provider + Dynamic
A Records (dns.ejs), Add New User + User List (users.ejs), Add
Permission + Permissions (permissions.ejs), Add Group (groups.ejs).
Bumps to v1.1.11.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KDEx8ghuZR61pqPXc6da9C
- A ? icon in the top-right header deep-links to the doc most relevant to
the current page (client-side path mapping, same pattern already used
for top-nav active-link highlighting -- no server-side "current section"
local exists to key off of instead). Falls back to the docs index.
- GET /docs/search does a plain line-substring search over the existing
allowlisted doc set. No new dependency, stays usable with no internet
access.
Bumps to v1.1.10.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KDEx8ghuZR61pqPXc6da9C
- Plain hosts can now be renamed after creation (wildcard/child/cache hosts
stay locked, since other records reference them by name). Migrates the
cert cache key on rename.
- Along the way, found and fixed a real bug in the vendored model-redis
library: its rename path leaves a stray, incomplete hash behind under
the old key when an `always`-type field (updated_on) is defined earlier
in the schema than the primary key -- silently blocking that hostname
from ever being reused. Worked around at the Host model level (can't
patch node_modules).
- Host list now shows who created each host, and when.
- Host modal's tabs now scroll horizontally on narrow screens instead of
overflowing awkwardly.
- Added missing inline help text (Target SSL, wildcard matching behavior).
Bumps to v1.1.9.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KDEx8ghuZR61pqPXc6da9C
- Host.prototype.update() had no challengeType handling (only create() did),
so selecting "Parent Wildcard" on an existing host's edit form silently
did nothing. Added the same wildcard-parent lookup to update(), using a
new Host.lookUpWildcardParent() -- the existing lookUp() can't be reused
here since an already-created host resolves to its own leaf rather than
falling through to a sibling wildcard.
- A wildcard's issued cert covers both the base domain and *.base domain
(altNames), but the lookup tree stores the wildcard one level below its
base -- looking up the bare base domain landed on an empty parent node
and found nothing. buildLookUpObj() now also stamps that parent node,
order-independent (a real host explicitly created at that exact name
always still wins).
Verified both fixes against a real Redis-backed Host model (not just the
mocked lookup-tree tests) -- see PR description.
Bumps to v1.1.8.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KDEx8ghuZR61pqPXc6da9C
The three auth_mode radios (Off / Basic / SSO) had no shared [name]
attribute, so per the HTML spec each was its own independent group --
clicking one didn't uncheck the others, letting multiple options
appear selected at once despite the page's own text saying "basic
auth and SSO can't both be enabled."
Added name="auth_mode" to restore native browser radio-group
behavior. The original comment claimed the radios were deliberately
kept nameless to avoid polluting the submitted form data (formAJAX
serializes every [name] field in the form), but that reasoning
doesn't hold: model-redis's processKeys() rebuilds the saved object
strictly from the Host model's own _keyMap, so an unrecognized
auth_mode field is silently stripped before anything is ever
persisted -- confirmed directly with model-redis's own
object_validate.js. Updated the stale comment accordingly.
jq-repeat 2.1.0 (release notes: https://github.com/wmantly/jq-repeat/releases/tag/v2.1.0)
brings real fixes (throttled-update race conditions, sorted-list
reverse() leaking elements, nested-scope isolation) and a few
behavior changes. Audited every usage in this repo against the
changelog before upgrading:
- push()/unshift() now return the new array length -- every call
site in this repo is a bare statement, none consume the return
value. No risk.
- update() is now trailing-edge throttled (~50ms) even on the first
call, not just rapid subsequent ones -- no code in this repo reads
DOM/item state immediately after calling update(), so no risk here
(unlike sso-manager-node's companion PR, which needed a fix).
- jr-order-reverse and nested jq-repeat templates: not used anywhere
in this repo.
Real breakage found and fixed: users.ejs/groups.ejs/permissions.ejs
called $.scope.X.__setPut(fn)/__setTake(fn) as setter METHODS -- that
API is gone in 2.1.0. Insert/remove hooks are now set via direct
property assignment ($.scope.X.__put = fn), per the current README.
Verified live (real dev server + Playwright): before the fix, all
three pages threw "__setTake is not a function" and the
insert/remove row animations were broken; after, zero errors and the
hooks fire correctly.
<title>, the navbar brand text, and the logo were all hardcoded
"Proxy - Theta 42"/"Dynamic Proxy". New conf.name/conf.logo keys
(defaults matching current text/asset) thread through the existing
values object pattern in routes/render.js and routes/docs.js, and
top.ejs now renders <%- name %>/<%- logo %> for the title and a new
navbar logo image.
Footer (copyright, theta42.com link, GitHub/license links) and the
existing favicon.svg are left as-is -- open-source attribution and a
distinct, already-working icon asset, not deployment branding.
GitHub Releases already carried real changelog notes per tag, but
those require internet access to view -- exactly what the /docs
route exists to avoid. CHANGELOG.md is a committed, Keep-a-Changelog
style file (backfilled from the v1.1.0/v1.1.1/v1.1.2 release notes),
linked from README and served at /docs/changelog alongside the rest
of the project's docs.
GitHub Releases already carried real changelog notes per tag, but
those require internet access to view -- exactly what the /docs
route exists to avoid. CHANGELOG.md is a committed, Keep-a-Changelog
style file (backfilled from the v1.1.0/v1.1.1/v1.1.2 release notes),
linked from README and served at /docs/changelog alongside the rest
of the project's docs.
Public route reading from disk on every request with no throttling
-- add a per-IP limiter matching the routes/auth.js/routes/host.js
convention already used elsewhere in this repo.
Air-gap:
- DynamicRecord.refreshAll() called getPublicIp() (api.ipify.org,
icanhazip.com, ifconfig.me) every 4h on a timer regardless of
whether any dynamic records were configured -- the one background
call in the repo not actually gated by feature use. Now skips the
lookup entirely when there's nothing to refresh.
- Removed the stray, unauthenticated GET /test page (a leftover
jq-repeat demo) that loaded jQuery + Mustache from external CDNs.
- Removed a dead IE<9-only html5shim script tag pointing at a domain
that no longer resolves.
Docs:
- New GET /docs (index) and /docs/:slug routes render this project's
own README, DEPLOYMENT, api.md, and docs/*.md server-side via
marked (new dependency) -- so the documentation is readable from
the running app with no route to GitHub Pages, where it otherwise
only lives. Public, no auth, same tier as the health endpoint.
- .dockerignore/Dockerfile updated: docs/, DEPLOYMENT.md, and
nodejs/api.md were previously excluded from the image entirely
("served via GitHub Pages, not from the image") -- now copied in
alongside README.md/tos.md-style, since they're needed at runtime.
listDomains() used to validate the token by calling DuckDNS's update
endpoint with ip/ipv6 omitted, which makes DuckDNS auto-detect and
apply this host's public IP -- so adding a provider instantly
repointed the domain. Validate via a fixed TXT marker instead, which
DuckDNS's API supports independently and doesn't touch routing.
ops/backup.sh snapshots Redis (BGSAVE, dynamic RDB path lookup) and
./config for standalone deployments, with retention. A background
service polls GitHub releases every 24h and surfaces an admin-only
banner in the UI when a newer version is published.
CodeQL flagged POST/PUT/DELETE /api/host* as missing rate limiting despite
performing authorization -- same authLimiter pattern routes/auth.js already
uses, applied here with a higher ceiling since legitimate admin work (bulk
edits) is expected on these routes.
CodeQL also flagged utils/basicauth.js's SHA-1 hashing as reachable from the
new basicauth-user route -- this is the existing, documented htpasswd-
compatible {SHA} scheme (see the comment on hashPassword), not something
this PR changes; left as-is per that comment's existing "follow-up" note,
since swapping it requires a coordinated change to
ops/nginx_conf/hostfeatures.lua's verification and a migration path for
already-stored hashes.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- ops/nginx_conf/targetinfo.lua's wildcard-subdomain lookup fallback used
classic LuaSocket (require("socket.unix")) instead of an OpenResty
cosocket. LuaSocket is blocking, and called from an nginx worker it
stalls the ENTIRE worker — every other in-flight connection on it — for
the round-trip to the Node app. Worse, the Node side never
newline-terminated its response, so the old blocking receive() only ever
returned via its read-timeout-then-partial-read fallback, meaning every
single cache-miss lookup paid a fixed timeout penalty while blocking the
whole worker. Replaced with an ngx.socket.tcp() cosocket (unix-domain via
"unix:/path", the only cosocket API this lua-nginx-module ships) and
newline-terminated the Node service's responses so receive() actually
completes instead of timing out. Verified against a live container:
previously this crashed OpenResty's Lua VM entirely
(ngx.socket.unix doesn't exist); fixed version resolves fresh wildcard
subdomains in ~2ms.
- Add gzip compression (`compression` middleware) and far-future
Cache-Control on static assets (7d for vendor libs under
/static-modules, 1h for the app's own /static JS/CSS, which isn't
cache-busted). The admin UI is a traditional multi-page app that loads
~13 separate vendor/app JS+CSS files on every full navigation; previously
none of them were compressed and Cache-Control was `max-age=0` (Express's
default), forcing a revalidation round-trip for every asset on every page
view.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- Auth tab is now a single choice (Off / Basic auth / SSO) instead of two
independent toggles that could both be on at once, which made it
ambiguous which gate actually protected a request. Enforced both in the
UI and server-side (POST/PUT), accounting for partial PUT updates against
the existing record.
- Add per-user basic-auth management (change password, delete) so an admin
no longer has to blow away and retype the whole user list to remove or
rotate one account.
- Fix: `Model.errors.ObjectValidateError(...)` is a constructor and was
being called without `new` everywhere in this codebase. Without `new`,
`this` inside it was the module's shared `errors` object (mutated in
place) and the call evaluated to `undefined` — so every
`throw Model.errors.ObjectValidateError(...)` actually threw `undefined`,
which Express's `next(undefined)` treats as "no error" and silently
falls through to the catch-all 404 handler. Every host/user/group/
permission/dns-provider validation error (bad hostname, bad IP, etc.) was
showing a confusing "Page not found" instead of the real message.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- Nav bar already showed the logged-in user's name linking to /profile
(cl-username); just removed the separate "Profile" and "API Tokens"
nav items now that Profile covers both.
- Merge api_tokens.ejs into profile.ejs as a section below the existing
profile card. /api-tokens 301-redirects to /profile.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
* 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).
permissions.ejs (6 columns), users.ejs (3 columns incl. an inline
password-change form), and profile.ejs's domain-permissions table had
no .table-responsive wrapper, so on narrow/mobile viewports they'd
either overflow the page horizontally or force zoomed-out, unreadable
text instead of scrolling within the table. hosts.ejs already had the
wrapper — these three didn't.
Verified: EJS compiles, npm test 192/192 pass, and fetched each route
from a running instance to confirm the wrapper is present in the
served HTML.
Two compounding bugs, both hit while adding a DuckDNS provider:
1. Domain.get() normalized every lookup via tldExtract before
checking Redis. model-redis' Table.create() stores the new record
under the literal key it's given, then calls this.get() internally
to return the created instance — so for any domain tldExtract
doesn't recognize as a shared/public suffix (e.g. a DuckDNS name
like "myhost.duckdns.org", which tldExtract naively normalizes to
"duckdns.org"), that final read-back always missed and create()
threw EntryNotFound, despite the record having just been written
successfully. In other words: no *.duckdns.org domain could ever
be created. Fixed by trying the exact string first and only
falling back to the tldExtract-normalized parent if no exact
record exists — preserving the original "look up an arbitrary
hostname, find the Domain that governs it" behavior for real
lookups, while fixing create()'s own read-back of what it just
wrote.
2. create() saves the DnsProvider row first, then calls
updateDomains() as a separate step. If updateDomains() throws —
e.g. a Domain collides with a stale/orphaned record left over from
an earlier failed attempt (exactly what bug 1 was silently
producing) — the already-saved provider row was never cleaned up,
leaving a broken, domain-less provider behind despite the API
returning an error. Fixed by wrapping updateDomains() in its own
try/catch and removing the provider on failure.
That fix has its own subtlety: `instance` (from super.create())
has its `domains` relation resolved by super.create()'s own
internal get() call, which runs BEFORE updateDomains() creates any
Domain rows — so instance.domains is permanently stale (always
empty), on both the success and failure paths. Removing `instance`
directly would delete the provider but silently leave behind
whatever domains updateDomains() did manage to create. Fixed by
re-fetching (this.get(instance.id)) before both the success return
and the failure-path remove(), so relations are current in both
cases — the returned/API-response instance and the rollback's
cascade-delete.
Manually verified against a live Redis (this project's test
philosophy explicitly excludes Redis-ORM-dependent tests from the
automated suite — see test/README.md "Philosophy"):
- A pre-existing orphaned Domain (simulating bug 1's fallout) now
produces an accurate "already exists" error instead of a confusing
EntryNotFound for the wrong (normalized) domain name, and the
failed create() leaves zero orphaned providers behind.
- A genuinely new *.duckdns.org domain now creates successfully,
correctly links to its provider, and is fully cascade-deleted when
the provider is removed.
- npm test: 192/192 pass.
Reported error when adding a DuckDNS provider with
subdomains="nl-theta42.duckdns.org" (the full name, as DuckDNS's own
site displays it):
{"name": "EntryNotFound", "message": "Domain:duckdns.org does not exists"}
listDomains() blindly appended ".duckdns.org" to whatever was entered,
turning "nl-theta42.duckdns.org" into
"nl-theta42.duckdns.org.duckdns.org". tld-extract doesn't know
duckdns.org is a shared suffix, so it parsed that malformed string
down to domain "duckdns.org" — surfacing as a confusing EntryNotFound
two layers away from the actual cause (Domain.create's internal
lookup).
Add __normalizeLabel() to strip a trailing ".duckdns.org" (and
lowercase) before use, so both "myhost" and "myhost.duckdns.org" work
identically. Also make __label()'s existing suffix-strip
case-insensitive to match.
wmantly